Clinical Features, Genome Epidemiology, and Antimicrobial Resistance Profiles of Aeromonas spp. Causing Human Infections: A Multicenter Prospective Cohort Study

Abstract Background The genus Aeromonas is increasingly implicated in human infections, but knowledge of its clinical characteristics and antimicrobial resistance profiles has been limited owing to its complex taxonomy. Methods We conducted a multicenter prospective cohort study of patients with Aeromonas infections at hospitals across Japan. Patients were eligible for inclusion if they had an Aeromonas spp. strain in a clinical culture and were considered infected at the culture site. Clinical data were collected, and isolates underwent susceptibility testing and whole-genome sequencing. Results A total of 144 patients were included. Hepatobiliary infection accounted for a majority of infections (73% [105 of 144]), which mostly occurred in elderly patients with comorbid conditions, including hepatobiliary complications. The all-cause 30-day mortality rate was 10.0% (95% confidence interval, 4.9%–14.8%). By whole-genome sequencing, 141 strains (98%) belonged to 4 Aeromonas species—A caviae, A hydrophila, A veronii, and A dhakensis—with significant intraspecies diversity. A caviae was predominant in all infection sites except skin and soft tissue, for which A hydrophila was the prevailing species. The genes encoding chromosomally mediated class B, C, and D β-lactamases were harbored by 92%–100% of the isolates in a species-specific manner, but they often lacked association with resistance phenotypes. The activity of cefepime was reliable. All isolates of A hydrophila and A dhakensis carried an mcr-3-like colistin resistance gene and showed reduced susceptibility to colistin. Conclusions Hepatobiliary tract was the most common infection site of Aeromonas spp., with A caviae being the dominant causative species. The resistance genotype and phenotype were often incongruent for β-lactam agents.

The genus Aeromonas consists of gram-negative facultative anaerobic bacilli that are ubiquitously distributed in aquatic environments.Aeromonas spp.have been increasingly recognized as important pathogens not only to fish and other poikilothermic animals but also to humans [1], and they are linked to numerous human infectious diseases, ranging from mild illness, such as enterocolitis, to life-threatening conditions, including necrotizing fasciitis [1,2].Despite the growing clinical significance of Aeromonas infections, our understanding of them has been hampered by difficulty in accurate species identification based on standard microbiological techniques, even with the use of matrix-assisted laser desorption ionization-time of flight mass spectrometry [3,4].Furthermore, a significant change in the nomenclature over the past 2 decades complicated species-level identification process of this taxonomic group, now consisting of at least 30 Aeromonas species, including A hydrophila, A caviae, A veronii, and A dhakensis, which are mesophilic Aeromonas spp.frequently associated with human infections [1,2,5].
The complexity in speciation has also compromised our understanding of antimicrobial resistance pattern of this genus.Some Aeromonas species are intrinsically resistant to clinically important β-lactam agents through the expression of ≥2 discrete chromosomally mediated β-lactamases, including class B metallo-β-lactamases, class C cephalosporinases (AmpC), Cohort Study of Aeromonas Infections • OFID • 1

Open Forum Infectious Diseases
and class D oxacillinases (OXA) [1,3].These β-lactamases are known to be produced in unique combinations in each species, with their expression regulated by the 2-component regulatory system BlrAB [6][7][8].However, species-level data on their prevalence and concordance with the resistance phenotype are still lacking.Furthermore, studies in recent years have suggested that some Aeromonas spp.harbor colistin resistance genes on their chromosome as potential sources of plasmid-mediated mcr-3 and mcr-7 in Enterobacterales [9][10][11].
We therefore aimed to investigate the clinical features of patients with diagnosed Aeromonas infections and to elucidate the genotypic and phenotypic characteristics of the causative bacterial strains, using whole-genome sequencing (WGS) data to support their taxonomic assignment.The species-specific antimicrobial resistance profiles were also evaluated, with the goal of informing treatment of patients with Aeromonas infections.

Study Design and Population
This was a prospective multicenter cohort study of individuals with Aeromonas infection recruited at 6 tertiary hospitals in Japan between June 2020 and August 2022.The hospitals included 3 general hospitals, 2 university hospitals, and 1 cancer center.Patients were eligible for inclusion if Aeromonas spp. was isolated from a clinical culture and they were considered infected at the culture site, defined by standard criteria (Supplementary Materials).The strain must also have been available for further analysis.Cases deemed to represent only colonization were excluded.There was no exclusion based on age.During the study period, only the first episode of Aeromonas infection was included for each patient.The study was approved by the institutional review board at Fujita Health University Hospital (no.HM20-162) and all other participating hospitals.The opt-out recruitment method was used, and the requirement for informed consent was waived.

Clinical Data Collection and Definition
The clinical data were collected from electronic medical records.For hepatobiliary infection, information on antimicrobial therapy was also collected.Detailed criteria and definitions used for each category are available in the Supplementary Materials.

Microbiological Analysis
The strains were initially identified as Aeromonas spp.by the standard microbiological procedure at each hospital and were subsequently transported to the central research laboratory for further testing.Antimicrobial susceptibility was evaluated by means of broth microdilution, using a customized 96-well plate (Eiken Chemical) in accordance with the manufacturer's instructions.The results were interpreted according to the Clinical and Laboratory Standards Institute document M45 [12].For antimicrobial agents without available breakpoints for Aeromonas spp. in that document, the 50th and 90th percentiles of the minimum inhibitory concentration (MIC) distributions (MIC 50 and MIC 90 ) and MIC ranges were presented.

Species Identification and Molecular Analysis
All study isolates were subjected to WGS.Species-level identification was performed by calculating average nucleotide identity values against the corresponding type strain genome (Supplementary Table 1), with a cutoff value of 95% used for species delineation, combined with a phylogenetic analysis based on core-genome single-nucleotide polymorphisms.The details of genome sequencing and assembly, species identification, phylogenetic analysis, multilocus sequencing typing, and identification of antimicrobial resistance genes are available in the Supplementary Materials.

Statistical Analysis
Categorical variables were compared with χ2 or Fisher exact tests, and continuous variables with Wilcoxon rank-sum test.Differences in the incidence of Aeromonas infections based on average monthly temperatures were evaluated using ordinal logistic regression models.Cumulative 14-and 30-day mortality rates were estimated using the Kaplan-Meier method.Detailed information on statistical analyses is available in the Supplementary Materials.

Patient Characteristics and Clinical Features
A total of 146 patients with Aeromonas infection were identified during the study period.After exclusion of 2 patients whose isolates were identified as species other than Aeromonas spp.by WGS analysis, 144 patients were included in the study.In this cohort, the incidence of Aeromonas infection was higher in months with an average monthly temperature >20°C (June-September) than in other months (odds ratio, 1.45 1. Given the high prevalence of hepatobiliary infection, univariate analysis was used to compare baseline characteristics between the patients with hepatobiliary infection and those with other infections.Those with hepatobiliary infection were more likely to have hepatobiliary complications (ie, biliary obstruction/stricture due to underlying cancer) than those without hepatobiliary infection (83% [87 of 105] vs 39% [15 of 39], respectively; P < .001)and had a higher median Charlson comorbidity index (interquartile range) (6 [5.0-8.0]vs 5 [3.0-7.0];P = .048).No significant difference was observed in other underlying conditions.
In terms of causative species, A caviae was the most common species, accounting for 60% (87 of 144), followed by A hydrophila (17% [25 of 144]), A veronii (14% [20 of 144]), and A dhakensis (6% [9 of 144]).These 4 species composed 98% (141 of 144) of the entire cohort, with the remaining strains identified as Aeromonas allosaccharophila (n = 2) or Aeromonas media (n = 1).Aeromonas spp.were coisolated with other pathogenic bacteria in 74% of cases, primarily residents of the enteric flora such as Escherichia coli, Klebsiella pneumoniae, and Enterococcus faecalis (Supplementary Table 2).A caviae was the predominant causative species in all infection types except skin and soft-tissue infections, from which A hydrophila was most frequently recovered.More specifically, A hydrophila was found to differ in the composition of infection sites (P = .04),with skin and soft-tissue infections accounting for a significantly higher proportion (Supplementary Table 3).

Genome Epidemiology of Aeromonas spp. Causing Human Infections
A core-genome single-nucleotide polymorphism-based maximum likelihood tree was generated with nonduplicate isolates obtained from 144 patients to evaluate the phylogenetic relatedness of the study strains (Figure 1).The tree consisted of 4 major clades, corresponding to A caviae, A hydrophila, A veronii, and A dhakensis.The strains belonged to 132 discrete sequence types (STs), of which 20 were known STs and 112 were novel STs.Four A caviae ST1825 strains were recovered from unrelated individuals at 3 hospitals.Otherwise, a majority of the isolates belonged to unique STs, reflecting a highly diverse population of Aeromonas strains recovered in this study.No specific lineage was associated with particular infection site, including the hepatobiliary tract.
Next, to assess the genetic relatedness of the study strains and those collected from human sources in different geographic locations, we conducted a phylogenetic analysis using the genomes obtained in this study and publicly available Aeromonas genomes of human origin, deposited in GenBank as of 7 November 2022 (Supplementary Table 4), for 4 major Aeromonas species.A high degree of genetic diversity was observed in Aeromonas strains recovered from human sources, and there was no clustering of genomes correlating with particular geographic regions (Supplementary Figure ).

Antimicrobial Susceptibilities
The antimicrobial susceptibility patterns of 4 major Aeromonas spp.are summarized in Table 2. Overall, cefepime, aztreonam, gentamicin, and trimethoprim-sulfamethoxazole demonstrated activity against 4 major Aeromonas species, with the susceptibility rates exceeding 89%.Fluoroquinolones were also active against a majority of the isolates, except for those belonging to A dhakensis, for which the susceptibility rate was slightly lower (78%).The susceptibility rates to β-lactam/β-lactamase inhibitors (eg, piperacillintazobactam), third-generation cephalosporins, and carbapenems were variable among the species, likely owing to the presence of chromosomally encoded β-lactamases, which are discussed below.Notably, the MICs of colistin were significantly higher for A hydrophila and A dhakensis, with MIC 50/ MIC 90 values of >4/>4 μg/mL, compared with other species (P < .001).

Antimicrobial Resistance Gene Profiles and Their Correlation With Resistance Phenotypes
The presence of antimicrobial resistance genes and the antimicrobial susceptibility testing results are shown in Figure 2. A gene that chromosomally encodes a class B2 metallo-β-lactamase which efficiently hydrolyzes carbapenems [13], bla CphA , was harbored by 92%-100% of the isolates of A hydrophila, A dhakensis, and A veronii, while the discrete AmpC genes, bla MOX , bla CepH / CepS, and bla AQU, were detected in 96%-100% of the isolates of A caviae, A hydrophila, and A dhakensis, respectively.The incidences of these antimicrobial resistance genes in individual species were similar to those reported in previous studies [14,15].Despite the high prevalence of these β-lactamase genes, their concordance rates to the phenotypic resistance (ie, nonsusceptible testing results) were variable among species, ranging from 53% to 100% for bla CphA and from 17% to 78% for bla AmpC (Supplementary Table 5).For the isolates carrying bla CphA, the resistance rate to imipenem was higher than that to meropenem, which was consistent with prior kinetic studies of CphA [16,17].Among 4 major species, A dhakensis showed the highest concordance for both bla CphA and bla AmpC , corresponding to high rates of carbapenem and cephalosporin resistance (Table 2).The bla OXA genes encoding OXA were present in all isolates in species-specific manner, and they were classified into 2 groups based on amino-acid sequence similarities: bla OXA-504-like (A caviae) or bla OXA-12-like (A hydrophila, A dhakensis, A veronii) [18].It could not be determined whether the OXA enzymes affected the susceptibility to β-lactam/ β-lactamase inhibitors [6,19].Genes encoding acquired β-lactamases, such as extended-spectrum β-lactamase or carbapenemase, were not detected in any of the study isolates.

DISCUSSION
Aeromonas spp. is increasingly recognized as a common human pathogen, but its population structure pertaining to human infections and their clinical implications are not well understood.The present study yielded several salient findings that serve to improve our understanding of this complex genus as a significant cause of invasive infections in humans.
We observed higher incidences of human Aeromonas infections during the warmer months of the year.This phenomenon, infection seasonality, has been well documented in mesophilic Aeromonas spp.and other gram-negative rods (eg, Acinetobacter spp.), which grow optimally at elevated temperatures [20,21].It is believed that the higher concentrations of these organisms in the environments during warmer months increase the likelihood of humans contracting them, although the details of the underlying mechanisms and transmission routes remain obscure [1,21].Hepatobiliary infection, particularly cholangitis, accounted for a majority of infections caused by Aeromonas spp.and tended to occur in elderly individuals with comorbid conditions, such as cancer and hepatobiliary complications.
These findings align with the epidemiological trend observed in East Asia, where a relatively high incidence of Aeromonas hepatobiliary infections has been reported, although not as high as observed in the present cohort (18% vs 73%, respectively) [15,22,23].On the other hand, there were only few cases of skin and soft-tissue infection or enterocolitis, which are classically considered as the major clinical manifestations of Aeromonas spp. in the United States, Europe, and Australia [20,24,25].Among the causative species, A caviae was predominant in most infection sites, including the hepatobiliary tract.The only exception was skin and soft tissues, where A hydrophila was isolated in significantly higher proportions, as described in the existing literature [15,20,26].Overall, genetically diverse Aeromonas strains were involved in human infections, without any specific lineages representing hepatobiliary infection or distributed in specific geographic locations [27,28].These findings suggest that the unique epidemiological features observed in this study cannot be explained by specific Aeromonas clones circulating in the community.Instead, they are more likely attributable to a composite of various interactions among the pathogen, host, social, and environmental factors.
Aeromonas spp.are not part of the normal enteric flora of humans; however, transient colonization occurs following the consumption of contaminated foods or drinking water, with the fecal colonization rate ranging from 1% to 30% in Patients in each group were divided according to high and low qSOFA scores, and no comparison between two groups was made for the low scores, because it was not clinically relevant.

Antimicrobial Agent
A caviae (n =
Cohort Study of Aeromonas Infections • OFID • 7 Japanese cuisine, characterized by frequent consumption of raw fish, is associated with increased risk of gastrointestinal colonization of Aeromonas spp.and, consequently, ascending bacterial infection in individuals with predisposing conditions.However, given the growing popularity of raw fish diet, the role of Aeromonas spp. as potential etiological agent of "foodborne illness" and their pathogenicity toward extraintestinal organs, particularly the hepatobiliary tract, may be speculated.
The genome-wide analysis of 144 clinical isolates also uncovered the species-specific distribution of antimicrobial resistance genes in Aeromonas spp., with chromosomal β-lactamase genes (bla CphA, bla AmpC , and bla OXA ) harbored by 92%-100% of the isolates within individual species.Nevertheless, their resistance genotypes and phenotypes were often incongruent, likely owing to variable expression of the enzymes [7,8].Prior studies suggested that additional tests (eg, susceptibility testing with a large inoculum, CarbaNP, or a modified carbapenem inactivation method (mCIM)) are required to detect CphA production [14,33,34].Interestingly, A dhakensis, increasingly recognized as a pathogenic species to be related to a high mortality rate in bacteremic patients [35], exhibited the highest genotypic and phenotypic concordance for both bla AmpC and bla CphA , corresponding to higher rates of carbapenem and cephalosporin resistance compared with other species.
Among β-lactam agents, the activity of cefepime was reliable in all species, which could be explained by its stability against both CphA and AmpC enzymes.Meanwhile, β-lactam/ β-lactamase inhibitors showed variable susceptibility patterns, without apparent association with the presence or absence of OXA and AmpC enzymes.Although the information is scarce on how these intrinsic resistance mechanisms may affect clinical outcome of the patients with Aeromonas infection [15], the genotypic and phenotypic discordance should be considered when interpreting the susceptibility testing results and selecting treatment options for Aeromonas spp.infection, especially for critically ill patients.
As of today, 10 plasmid-mediated mcr genes (ie, mcr-1 to mcr-10) have been described in various gram-negative bacilli worldwide, conferring reduced susceptibility to colistin [36].Among them, mcr-3 and mcr-7 have been speculated to originate from Aeromonas spp., based on the presence of the Aeromonas chromosomal gene sequences that align closely with these mcr genes [9,10].However there have been conflicting data on the prevalence of mcr in individual Aeromonas spp.[11,37], and the information on their colistin resistance has been scarce [11,38].In our study mcr-3-like genes were harbored in all isolates of A hydrophila and A dhakensis, with their presence correlating with high MICs of colistin, suggesting that these 2 species are intrinsically resistant to colistin.In contrast, mcr-7-like genes, detected in a majority of the A veronii strains, did not confer colistin resistance.This is in line with prior studies showing poor ability of MCR-7 to elicit colistin resistance, likely due to structural differences of the enzymes [39].According to these findings, colistin is not recommended for the treatment of infection suspected or confirmed to be caused by A hydrophila and A dhakensis.Our study has several limitations.First, it was a multicenter study conducted in Japan, and may not reflect trends in other geographic locations.Second, the limited number of A dhakensis isolates may have resulted in missed characteristics.Third, the majority of Aeromonas infections were polymicrobial, where the pathogenic role of each organism was difficult to be determined.Fourth, the clinical impact of infections caused by a strain exhibiting incongruent phenotypic and genotypic resistance could not be assessed owing to the small number of cases with outcomes.Fifth, WGS used for species identification in this study is still difficult to implement in general clinical settings.
In summary, hepatobiliary infection was the most common clinical manifestation of human Aeromonas infections, with A caviae being the predominant causative species.The prevalence of antimicrobial resistance genes was species specific, and resistance genotype and phenotype were often incongruent for β-lactams, which should be considered when interpreting the susceptibility testing results and selecting antimicrobial agents for therapy.

Table 3 . Treatment and Clinical Outcome of Patients With Aeromonas Infections, Hepatobiliary or Nonhepatobiliary
Abbreviations: CI, confidence interval; ICU, intensive care unit.a Data represent no.(%) of patients unless otherwise specified.
Phylogenetic tree of 144 Aeromonas isolates from Figure1shown with the susceptibility testing results and antimicrobial resistance genes.Nonsusceptibility testing results for carbapenems (imipenem and meropenem), third-generation cephalosporins (cefotaxime and ceftazidime), cefepime, and piperacillin-tazobactam are shown with orange, light blue, blue, and light gray boxes, respectively.For colistin, minimum inhibitory concentration (MIC) values are displayed with a color gradient, from white (low MICs) to green (high MICs).The presence of resistance genes is noted with colored boxes; orange represents bla CphA ; blue, bla MOX , bla CepH / CepS, bla AQU , and bla FOX ; gray, bla OXA-504-like, bla OXA-12-like, and bla OXA-917 ; green, and mcr-like gene.