The Aeromonas species are Gram-negative, rod-shaped bacteria that inhabited soil and aquatic environment ubiquitously, from fresh and brackish water, seawater, groundwater, sewage to drinking water. In addition, they were also found in fish and seafood, dairy, meats, and vegetables intended for human consumption (Fernández-Bravo and Figueras, 2020). They cause a wide spectrum of diseases in humans, notably acute gastroenteritis, septicemia, and soft tissue infections, as well as hepatobiliary tract infections, peritonitis, respiratory tract infections, urogenital tract infections, indwelling-device related infections, ocular infections, meningitis, and hemolytic uremic syndrome (Janda and Abbott, 2010).

The pathogenicity of aeromonads is complex owing to their multiple virulence factors acting collectively or separately, including structural components like flagella, adhesins, lipopolysaccharide and capsule, extracellular enzymes like lipases, proteases, elastases, and hemolytic enzymes that cause cell and tissue damage, enterotoxins that induce diarrhea, and most importantly the type III secretion system (T3SS) that injects toxins directly into host cells. The T3SS is composed of thorn-shaped or syringe structure, effector proteins that are injected, and chaperones that assist and protect structural and effector proteins during transport (Tomás, 2012; Rasmussen-Ivey et al., 2016; Gonçalves Pessoa et al., 2019). AscV serves as an indicator for the presence of the type III secretion machinery. AscF-AscG serves as translocation apparatus (Vilches et al., 2004; Chacón et al., 2004; Burr and Frey, 2007). AexT is an effector protein possessing ADP-ribosyltransferase and GTPase acting protein activities and is homologous to the Pseudomonas aeruginosa effector ExoT/ExoS (Braun et al., 2002; Tomás, 2012). The cytotoxic enterotoxin Act provokes the degeneration of intestinal epithelium and leads to bloody diarrhea, while the cytotonic enterotoxins, including heat-stable type Ast and heat-labile type Alt, cause non-bloody diarrhea (Gonçalves Pessoa et al., 2019).

Southern Taiwan locates in a subtropical area and is an Aeromonas-prevalent region, with an incidence of Aeromonas bacteremia of 76 per million inhabitants per year, much higher than that in western countries with an annual incidence of merely up to 1.5 per million (Wu et al., 2014). Historically, A. hydrophila had been the most common species isolated in bacteremia in Southern Taiwan (Ko and Chuang, 1995; Ko et al., 2000; Tang et al., 2014), but recent advances in molecular studies based on 16s RNA (Martínez-Murcia et al., 1992), housekeeping genes (Yáñez et al., 2003; Soler et al., 2004; Martínez-Murcia et al., 2011), and genome sequencing (Colston et al., 2014), had led to the reclassification of aeromonads. As a result, the reported prevalence of the most predominant clinical species of Aeromonas has changed over the years, with most (96.5%) of the aeromonads associated with clinical cases identified as A. caviae (29.9%), A. dhakensis (26.3%), A. veronii (24.8%), and A. hydrophila (15.5%) (Fernández-Bravo and Figueras, 2020). Besides, concordance was low between phylogenetic identification and the commercial identification systems, with incorrect identification at species level (Lamy et al., 2010). For example, it could be difficult to separate A. veronii biovar sobria from A. hydrophila using conventional biochemical tests. A. veronii biovar sobria shares common phenotypes with A. sobria sensu stricto and was often reported mistakenly as A. sobria (Janda and Abbott, 2010). A. dhakensis was mistaken as A. hydrophila for decades and is often misidentified as A. hydrophila, A. veronii, or A. caviae by commercial phenotypic tests (Chen et al., 2016). Since 16s RNA is highly conserved in aeromonads, housekeeping genes like gyrB (subunit B of DNA gyrase) and rpoD (sigma factor S70) offer less mean sequence similarity values and hence higher resolution than the 16s RNA gene (Yáñez et al., 2003; Soler et al., 2004; Martínez-Murcia et al., 2011).

In this prospective study, we investigated patients with clinical isolates of Aeromonas species determined by DNA sequence matching of rpoD or gyrB between 2016 to 2018 in a medical center in Southern Taiwan. The demographic factors, clinical outcome, drugs susceptibility of Aeromonas isolates, and the prevalence of genes responsible for drug resistance and virulence were analyzed. The study aimed to provide a better understanding of the association between clinical spectrum and different Aeromonas species determined by molecular typing.

The prevalence of human infections caused by A. veronii and A. dhakensis might be underestimated since both would be misidentified as A. hydrophila or A. sobria by the phenotype-based identification system or even MALDI-TOF MS as shown in the present study. A. dhakensis was found to be the most common Aeromonas species isolated from wound cultures, more virulent than A. hydrophila ex vivo and in animal models (Chen et al., 2014), as well as harboring the highest 14-day sepsis-related mortality rate among monomicrobial Aeromonas bacteremia (Wu et al., 2015). A. dhakensis was found to be the dominant aeromonad in Singapore and Malaysia, accounting for 45-50% of all Aeromonas species identified (Puthucheary et al., 2012; Khor et al., 2018). In Australia, A. dhakensis was the most prevalent aeromonad in clinical and water samples, especially in wounds (Aravena-Román et al., 2011). T3SSs are found in many Gram-negative bacterial pathogens including Pseudomonas, Yersinia, Salmonella, Shigella, as well as enteropathogenic and enterohemorrhagic Escherichia coli (Wu et al., 2007). The T3SS of Aeromonas is similar to that of Yersinia (Vilches et al., 2004), with at least 21 effector proteins (Rangel et al., 2019) that exhibit cytotoxicity, induce apoptosis, reduce phagocytosis, and trigger cytokines/chemokines production (Yu et al., 2004; Burr et al., 2005; Sierra et al., 2010). Strains of A. salmonicida and A. hydrophila with mutations in the T3SS apparatus were shown to be less virulent than non-mutated strains (Vilches et al., 2004; Yu et al., 2004; Burr et al., 2005). Our previous research demonstrated that ascF-ascG was mainly present in A. hydrophila, A. dhakensis, and A. veronii (50%, 14.3%, and 1%, respectively) (Wu et al., 2019), and ascV was previously more common in A. hydrophila comparing with A. dhakensis (92.3% vs 51.4%, p=0.017) (Chen et al., 2014), but there were shreds of evidence demonstrating an association between the presence of ascV, aexT or ascF-ascG genes and the development of extraintestinal infections or bacteremia among patients with Aeromonas isolates (Wu et al., 2007). The present study illustrated that ascV carriage was similar between A. hydrophila and A. dhakensis (13.7% vs 13.5%), and A. veronii possessed ascF-ascG gene more often than previously reported. The ascF-ascG gene was independently correlated to crude in-hospital mortality in the present study, a correlation that had not yet been elucidated in other studies.

The distribution of the AmpC β-lactamases and MBL genes were found to be species-specific in a previous study conducted in our hospital, with all A. dhakensis, A. caviae, and A. hydrophila isolates carrying bla_AQU-1 , bla_MOX , and bla_CepH , respectively (Wu et al., 2015). Consistent with this finding, the present study demonstrated increased resistance to third-generation cephalosporins among the three aeromonads harboring genes encoding AmpC β-lactamases. In the present study, bla_AQU-1 was found exclusively but not universally in 91.9% of the A. dhakensis isolates. Reduced susceptibility to cefepime was found among ESBL genes-carrying A. caviae and A. dhakensis isolates, but 2% of A. hydrophila isolates also exhibited cefepime resistance without identifiable ESBL genes in the present study. Resistance to ertapenem was quite high among aeromonads carrying the MBL gene bla_CphA , such as A. dhakensis, A. hydrophila, and A. veronii, and resistance to imipenem could be found in the aforementioned aeromonads, as well as bla_NDM -carrying A. caviae. A. caviae was found to carry bla_NDM on the chromosome from water seepage samples in New Dehli in 2010 (Walsh et al., 2011). Clinicians should be aware of the emergence of bla_NDM in A. caviae. Moreover, 57.1% of A. sobria isolates showed intermediate susceptibility to ertapenem without carrying bla_CphA or other carbapenemases tested in the present study. Other carbapenemases, such as the class D penicillinase AmpS, had been discovered in A. sobria (Walsh et al., 1995a; Walsh et al., 1995b). The two A. sanarellii isolates, one of them carrying bla_MOX gene, displayed non-susceptibility to cefotaxime and piperacillin/tazobactam, and one of them was resistant to ertapenem and tetracycline. Other AmpC β-lactamases and MBL or porin alterations not examined in this study may contribute to the drug resistance.

Aeromonas infection had been linked to patients with liver cirrhosis or cancer with poorer outcomes in Taiwan (Ko and Chuang, 1995; Ko et al., 2000; Wang et al., 2009), an island that had been endemic with hepatitis B (Chan et al., 2004). The present study demonstrated that the proportion of patients with active malignancy had surpassed liver cirrhosis as the most common underlying disease in patients with Aeromonas infection, possibly attributed to the mass vaccination program of hepatitis B vaccine since 1984, thereby reducing the carrier rate by 85% (Chan et al., 2004). Hepatocellular carcinoma (HCC), which was linked to liver cirrhosis and chronic hepatitis B and C infection, was the most common cancer type in the present study, accounting for 19.4% of all patients with active malignancies, and had the highest crude in-hospital mortality rate (6/14, 42.9%) among all comorbidities. Liver cirrhosis confers susceptibility to infection by immune dysfunction including reduced secretory IgA and bile acid, and alteration in the gut microbiome, making the host susceptible to infections originating from the gut (Bajaj et al., 2021). On the other hand, chemotherapies directed against malignancies confer susceptibility to food-borne infections by disrupting the gut mucosal barrier, and the following neutropenia predisposes the host to opportunistic infections (National Comprehensive Cancer Network, 2020). The risk of infection in patients with HCC and pancreatobiliary cancers may also be increased due to hepatobiliary obstruction caused by tumors, as a sequela of hepatobiliary reconstruction surgery (National Comprehensive Cancer Network, 2020), and resistance to bile salts of aeromonads (Want and Millership, 1990). Chao et al. discovered that patients with cancer are associated with higher mortality in Aeromonas bacteremia, pneumonia, and biliary tract infection (Chao et al., 2013a; Chao et al., 2013b; Tang et al., 2014). In contrast, the outcome of skin and soft tissue infection attributed to Aeromonas was associated with diabetes mellitus but not immune status (Chao et al., 2013c).

In the present study, patients with hematologic malignancies were infected by A. veronii and A. hydrophila exclusively, and the crude in-hospital mortality was 30.8% (4/13). In another tertiary medical center in Southern Taiwan, Tsai et al. found that 35.6% of 41 patients with hematologic dyscrasias succumbed to Aeromonas bacteremia within 14 days of onset, with a remarkably high resistant rate to imipenem (35.6%) (Tsai et al., 2006). Patients with hematological malignancies were particularly vulnerable to opportunistic infection owing to the frequent leukopenia due to marrow infiltration of malignant cells or dysfunctional marrow, the severe mucosal damage, and prolonged neutropenia following higher-intensity chemotherapies, and a shift in enteric microbial flora accompanied by severe illness and antimicrobial usage (National Comprehensive Cancer Network, 2020).

Biliary tract infection with aeromonads was associated with biliary stone and pancreatobiliary cancer in our study, agreeing with other studies in southern Taiwan and Japan (Chao et al., 2013a; Kitagawa et al., 2020). Pancreatic cancer was the second most common cancer in our study, and the sum of patients with pancreatobiliary cancers outnumbered patients with HCC. In Japan, pancreatobiliary cancer, liver cirrhosis, and obstructive biliary disease contributed equally to comorbidity in patients with Aeromonas bacteremia, and 57.9% of bacteremia originated from biliary tract infection (Kitagawa et al., 2020). The incidence of pancreatobiliary cancers was around 20.93 per 100,000 inhabitants per year in Taiwan, far way behind the incidence of breast cancer, colorectal cancer, lung cancer, prostate cancer, and hepatocellular carcinoma (119.71, 70.05, 65.05, 56.72, and 36.62 per 100,000 inhabitants per year, respectively) (Health Promotion Administration, Ministry of Health and Welfare, 2020). This phenomenon may be attributed to the preponderance and even possible carcinogenesis of aeromonads in patients with pancreatobiliary cancers, or simply an institutional bias since our hospital was renowned for the treatment of pancreatic cancer (Su et al., 2020). Overexpression of the p38 MAPK pathway is observed in the pancreatic cancer cells and hepatocellular cells (Dhillon et al., 2007). Our previous research (Chen et al., 2018) have discovered that A. dhakensis infection causes p38 mitogen-activated protein kinase (MAPK) pathway activation in Caenorhabditis elegans model. We presume that Aeromonas species living in the hepatobiliary tract trigger the development of cancers by activating p38 MAPK pathway. Further studies are warranted to clarify the causality between Aeromonas and pancreatobiliary cancer. Half (8/16) of the patients with biliary stones were infected with A. caviae and half (6/12) of those with pancreatic cancer with A. veronii. Since A. caviae and A. veronii carry bla_MOX and bla_CphA , respectively, cefepime and fluoroquinolone are drugs of choice in these patients.

Three-fourths of the patients with Aeromonas pneumonia died in the hospital. Pneumonia was the most significant factor associated with crude in-hospital mortality in the present study, yet it was caused by A. caviae, the least virulent species with the highest resistance to third-generation cephalosporins and therefore most healthcare-associated among the commonly encountered aeromonads (Wu et al., 2015). A study conducted in another hospital located in the same city addressing 84 patients with Aeromonas pneumonia showed that the in-hospital mortality was merely 28.6% (Chao et al., 2013c). Since the majority (10/13, 76.9%) of these patients had a polymicrobial infection, the high mortality in the present may reflect the complex comorbidities and prolonged hospitalization, not the virulence of Aeromonas itself.

This study had several limitations. First, clinical information was collected retrospectively. Selection bias may be present since there could be more severe patients and more patients with rarer malignancies such as pancreatobiliary carcinomas were referred to the study hospital. Second, this study was conducted in a single medical center, and a multicenter study is warranted for a more comprehensive understanding of the epidemiology of clinical infections caused by aeromonads in other areas. Finally, a severity score was not available in this study, which may contribute to the risk of mortality in the outcome analysis.

A. veronii, A. caviae, A. hydrophila, and A. dhakensis were the most frequently isolated species in Aeromonas infections. Infection with ascF-ascG Aeromonas and underlying malignancies were associated with mortality. Cefepime, fluoroquinolones, and tigecycline are the drugs of choice for Aeromonas infections, especially for skin and soft tissue infections and biliary tract infections in patients with underlying pancreaticobiliary cancers.

P-LC conceived and designed the experiments. S-LS performed all the experiments. Y-WC analyzed the data and drafted the paper. C-WL, N-YL, and C-ST provided technical help on data analysis. C-CL, W-CK, C-LL, L-SS, and M-CL critically commented on the analysis. P-LC reviewed and edited the paper. All authors contributed to the article and approved the submitted version.

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

The study was partially supported by research grants from the Ministry of Science and Technology, Taiwan (MOST 110-2628-B-006-028), and National Cheng Kung University Hospital (NCKUH-10902001 and NCKUH-110040027).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.