This study aimed to evaluate the epidemiology of carbapenemase-producing Escherichia coli and Klebsiella pneumoniae recovered over a three-year period (2020–2022) from four healthcare institutions in Lima, Peru, revealing the emergence and multiclonal dissemination of these pathogens. The spread of CPE is now widely established across Latin America and is frequently associated with the co-occurrence of resistance mechanisms to multiple antimicrobial classes, which significantly complicates clinical management of infections (21).

All carbapenemase-producing isolates in this study displayed a MDR phenotype, underscoring the potential limitations in available therapeutic options and highlighting the urgent need for a coordinated global response to address this emerging threat. Among the agents tested, ciprofloxacin, trimethoprim/sulfamethoxazole, and amikacin exhibited the lowest activity, with resistance rates exceeding 75%, and reaching 100% for ciprofloxacin. These findings confirm the limited empirical utility of fluoroquinolones and trimethoprim/sulfamethoxazole for the management of urinary tract infections caused by CPE, in line with the Infectious Diseases Society of America (IDSA) recommendations, which advise against the empirical use of these agents in settings where local resistance rates surpass 20% (22).

Drugs commonly employed in combination therapies for CPE infections showed variable activity against contemporary isolates. In our cohort, fosfomycin exhibited low resistance rates (<15%; 14.3% in K. pneumoniae and 6.6% in E. coli), comparable to data reported from Argentina (11.4%) (23), Mexico (10.9%), and China (10%) (24, 25). However, these results contrast with previous reports from Peruvian hospitals, where fosfomycin resistance reached 27.8% (26), and with studies from Southeast Asia and North Africa documenting prevalence up to 36.4% in Vietnam and 38.5% in Egypt (27, 28).

Regarding colistin, resistance rates reached 29.3% in K. pneumoniae and 8.2% in E. coli. Although colistin remains a last-resort option, its efficacy has been increasingly compromised by the rising prevalence of resistant strains across Latin America (29). A recent multicenter study in Peru analyzing 317 clinical isolates from 12 regions reported lower colistin resistance rates—7.5% in E. coli and 11.9% in K. pneumoniae (30)—values below those observed in the present study. This difference may be explained by our focus on carbapenemase-producing isolates, which have been reported to exhibit colistin resistance rates as high as 31% (31).

Given the limited therapeutic options available for severe infections caused by CPE, combination regimens including colistin and fosfomycin may provide valuable alternatives against these multidrug-resistant clinical isolates (32).

The high rate of amikacin resistance observed in this study is particularly concerning, given that this aminoglycoside has traditionally been used in Peru as a reserve agent against multidrug-resistant (MDR) Gram-negative bacilli and as an alternative therapeutic option in settings where access to newer antimicrobial agents remains limited. In E. coli, resistance to amikacin was higher than to gentamicin (75.4% vs 47.5%), whereas in K. pneumoniae, the opposite pattern was observed, with gentamicin showing higher resistance than amikacin (89.2% vs 65.6%). This discrepancy may reflect the presence of distinct aminoglycoside resistance genes conferring selective resistance to each compound (32).

Recent studies recommend the primary use of novel β-lactam/β-lactamase inhibitor combinations such as ceftazidime/avibactam (CZA) for infections caused by serine carbapenemase producers, and aztreonam/avibactam (AZA) or cefiderocol for those mediated by metallo-β-lactamases (MBLs) (32). In agreement with these recommendations, our study demonstrated complete susceptibility (100%) to CZA among serine carbapenemase producers, consistent with data from recent global surveillance programs that included isolates from Latin America (33, 34). However, although no resistance to CZA was detected among serine carbapenemase producers in our collection, reports from Peru have described CZA-resistant Klebsiella pneumoniae associated with KPC variants such as KPC-35, highlighting the potential for emerging resistance even within this group (101).

Aztreonam/avibactam exhibited the broadest activity spectrum, showing full activity against all carbapenemase classes and combinations thereof. These findings are consistent with prior studies demonstrating 100% susceptibility of AZA against MBL, serine carbapenemase, and co-producing isolates (35, 36, 37). Such results underscore the therapeutic potential of AZA as a key agent for managing MDR infections, particularly those mediated by MBL-producing Enterobacterales (38). This is especially relevant in Peru, where NDM-type carbapenemases remain the most prevalent (4).

Cefiderocol represents a valuable second-line option against NDM and other MBL-producing Enterobacterales, with evidence supporting high clinical efficacy (39). In our study, all isolates—including co-producers—remained susceptible to cefiderocol (100%). Nevertheless, further research is warranted, as a recent meta-analysis identified multiple β-lactamases and resistance determinants associated with increased cefiderocol MICs, particularly among NDM-producing strains (40).

This study confirms that NDM-type MBLs remain the most prevalent carbapenemases in Peru, surpassing KPC variants (4). These findings position Peru among the countries where NDM represents the dominant carbapenemase, in line with global trends reporting its dissemination across all continents (41, 42).

In contrast, this pattern is not universal. Surveillance reports from Europe, Latin America, and the Caribbean indicate that KPC-type carbapenemases remain the most widespread among Enterobacterales in those regions, having reached endemic levels in several countries (41, 43, 44).

Historically, K. pneumoniae has been the main reservoir for carbapenemases, as reflected in this study (21). However, E. coli has shown a progressive increase in the dissemination of carbapenemase genes, particularly since 2015 (43). This observation mirrors the global trend of expanding MBL-type enzymes into species already associated with carbapenemase propagation (21, 43).

A distinctive feature of our region—also evident in this study—is the low prevalence of OXA-48-like carbapenemases (41). In our cohort, OXA-48-like enzymes accounted for 4.6% of isolates, showing only a modest increase compared with previous surveillance periods (4), ranking as the third most frequent carbapenemase detected among Enterobacterales in Peru. Interestingly, within E. coli, OXA-48-like was the second most common enzyme (31.1% in E. coli vs. 5.4% in K. pneumoniae).

An additional emerging concern is the increasing detection of co-producing carbapenemase isolates since the onset of the COVID-19 pandemic. E. coli, previously regarded as an occasional reservoir, has begun to harbor multiple carbapenemase genes more frequently (21). Our findings confirm the persistent circulation of K. pneumoniae co-producing KPC and NDM (3.1%) after the first sporadic case reported in 2016 (45), with a notable rise during the pandemic period (46). Moreover, the first E. coli isolate co-harboring bla_NDM and bla_OXA-48-like was reported in 2021, followed by subsequent detections with similar genetic configurations (47).

A recent review suggests that the emergence of multi-carbapenemase producers may be linked to the accumulation of non-β-lactam resistance mechanisms, likely driven by selective pressure from other antimicrobial classes (48).

Whole-genome sequencing (WGS) enabled the characterization of the allelic variants of carbapenemase genes previously identified by PCR. The most frequent variant detected was bla_NDM-1 consistent with multiple reports from Peru (49, 50, 51). The bla_KPC-2 variant, also previously documented in the country (52), and bla_OXA-48-, first reported in Peru in 2022 (53), were likewise identified. These variants represent the most commonly reported carbapenemase alleles across Latin America (41, 43).

Notably, bla_OXA-181 was also detected—this variant has been recently reported in Peru among K. pneumoniae, E. coli, and Citrobacter spp. isolates (54), suggesting that its presence in the country is not sporadic but rather indicative of ongoing local dissemination. Additionally, bla_NDM-5 was identified; this allele has been described in Argentina, Uruguay, Brazil, Mexico, United States, Europe and Middle east (55, 56, 57, 58, 59, 60, 61, 62). To our knowledge, this represents the first report of bla_NDM-5 in Peru. Furthermore, co-producing isolates harboring bla_KPC-2 + bla_NDM-1 and bla_NDM-1 + bla_OXA-48 were detected, a phenomenon considered a global public health emergency (48).

Bioinformatic analysis of WGS data provided additional genomic insights into E. coli and K. pneumoniae isolates. MLST revealed two predominant E. coli clones: ST410 and ST167, which together accounted for 78.3% of sequenced E. coli isolates. E. coli ST410, belonging to clonal complex CC10, has been increasingly associated with the global spread of bla_OXA-181 (48) and is now recognized as one of the most prevalent carbapenem-resistant E. coli (CRE) lineages worldwide, with presence across all five continents (63). In this study, the fimH24 variant of E. coli ST410 was the second most frequent sequence type (30%, 7/23), strongly associated with chromosomally located bla_OXA-48. This finding is clinically relevant, as the presence of a single chromosomal copy of bla_OXA-48 may hinder detection by standard phenotypic methods, posing an additional diagnostic challenge for clinical laboratories.

Regarding E. coli ST167, this lineage is among the most globally distributed and has shown increasing occurrence in Latin America. It is frequently associated with bla_NDM-5, typically carried on IncF plasmids (63). In our study, ST167 was the predominant lineage among E. coli-CRE isolates (48%, 11/23), linked to the O101:H10 serotype and bla_NDM-5 carriage. However, unlike previous reports, our isolates lacked the fimH54 allele, previously proposed as a conserved feature in this lineage conferring enhanced colonization and evolutionary advantages (64). This pattern supports the local spread of a closely related clonal group (6–21 SNPs apart) across several healthcare institutions in Peru.

Among K. pneumoniae, four major STs were identified—ST147, ST273, ST15, and ST45—representing 58.3% of sequenced isolates. Multidrug-resistant K. pneumoniae is recognized as a critical global health threat (1) and its rapid dissemination is largely attributed to the geographic expansion of successful clonal groups (CGs) such as CG15, CG101, CG147, CG258, and CG307 (65, 66). K. pneumoniae ST147 has been identified as a globally distributed high-risk clone (67), closely related to ST273 and ST392, all of which belong to CG147 (68, 69). These clones commonly harbor fluoroquinolone-resistance mutations (gyrA S83I, parC S80I).

In our isolates, ST147 carried capsular loci KL64 (except one with KL20), while ST273 carried KL74. Historically, CG147 (primarily ST147) has been associated with multiple carbapenemase types since its emergence in the early 1990s, with ST273 first reported in 1995 (67, 70). Between 2010 and 2014, CG147 clones were reported globally in association with KPC, NDM, and OXA-48-like enzymes (71). In Peru, ST147 has been repeatedly identified and is emerging as a dominant clone responsible for carbapenemase dissemination in the country (72, 73, 74, 75).

K. pneumoniae ST15 is an emerging high-risk clone frequently implicated in hospital outbreaks (76). ST15 has been reported with multiple carbapenemases—including OXA-232, KPC-2, and NDM—in various settings (77) and has also been documented in Peru carrying NDM (78). In our series, ST15 isolates harbored QRDR mutations (gyrA S83I/D87A; parC S80I), the predominant capsular locus KL48, and the ybt1/ICEKp4 virulence module linked to siderophore-mediated iron acquisition (79, 80, 81).

K. pneumoniae ST45 has been described worldwide as a carrier of ESBLs and carbapenemases (82, 83, 84) and was recently reported in Peru as an NDM-producing lineage (78). In our 2020–2022 collection, ST45 isolates co-produced KPC-2 and NDM-1, a concerning pattern that intensified during the COVID-19 pandemic. These ST45 isolates lacked QRDR mutations, carried KL62, and encoded ybt1/ICEKp4, mirroring the virulence profile of our ST15 isolates.

Virulence in K. pneumoniae is associated with additional siderophores (ybt, iuc, iro) and specific capsular serotypes (K1, K2, K5) (85). Most hvKP infections arise in the Asia-Pacific region and are enriched in STs such as ST23, ST65, and ST86 (86). We identified one ST23 isolate carrying hallmark hvKP determinants (e.g., iro, clb) and rmpA associated with the hypermucoviscosity regulon. Importantly, hypermucoviscosity and hypervirulence are not synonymous—neither phenotype implies the other (87). Capsule overproduction (and thus mucoviscosity) can be modulated by magA, rmpA, rmpA2, and the two-component system rcsAB (87). Consistent with hvKP, our isolate belonged to capsular type K1 (86).