A 77-year-old male, with underlying coronary artery obstructive disease and diseases of idiopathic pulmonary fibrosis, hypertension, diabetes, and dyslipidemia, visited the emergency center presenting with left hand pain. He was administered aspirin, trim-etazidine, sacubitril/valsartan, torasemide, neustatin, metformin, dapagliflozin, lansoprazole, theobromine, bromhexine, and lafutidine at our hospital. Three days prior to visiting the emergency center, he fell and his left palm was pricked by a beanstalk. A day later, he visited a local hospital for treatment and was administered cefazolin intravenously. Computed tomography (CT) was performed at a local hospital and infectious tenosynovitis of the left hand was suspected. He was then referred to our hospital for emergency surgery. Approximately 2 cm of linear abrasion with pus and sanguineous discharge were observed in the middle of the left palm (Fig. 1). The left hand was diffusely swollen, and the patient complained of heating sensation, tenderness, and pain. No neurological deficits or blood circulation problems were found. Vital signs were 103/69 mmHg for systolic/diastolic blood pressure, 118/min for pulse rate, 18/min for respiratory rate, and 36.8°C for body temperature. The routine laboratory test detected leukocyte count of 10,820/mm³ with 72.7% of neutrophils, erythrocyte sedimentation rate of 26 mm/h (reference interval [RI], 0.0–22.0 mm/h), C-reactive protein of 119.4 mg/L (RI, 0–8 mg/L), and procalcitonin of 0.14 ng/mL (RI, 0.00–0.50 ng/mL). Typical symptoms and laboratory findings strongly suggested infectious tenosynovitis. The patient was referred to another hospital because of emergency surgery difficulties after collecting the wound discharge for microbiological investigations.

The white blood cells (WBCs) were 3+ in gram staining of wound discharge, but no organism was seen by microscopy. However, few medium-sized, colorless, smooth, convex, and glistening colonies were observed on blood agar and few medium-sized, pink, smooth, convex, punctate, umbilicated, and glistening colonies were observed on MacConkey agar plates incubated at 35°C under 5% CO₂ for 24 hours (Fig. 2). The microorganisms were medium-to-long, plump, gram-negative rods (Fig. 3) and were oxidase-negative and catalase-positive. In the triple sugar iron medium, the slant and the butt were yellow, and gas generation was confirmed, but hydrogen sulfide was absent. The Bruker Biotyper (Bruker Daltonics, Bremen, Germany) identified the microorganism as Kosakonia cowanii with a score of 2.12, which is reliable for species level identification according to the manufacturer’s instruction. As K. cowanii is a rarely identified bacteria in clinical laboratories, the VITEK 2 system (bioMérieux, Marcy l’Étoile, France) and VITEK MS (bioMérieux) were additionally used for confirming the results of bacterial identification. However, the former was unable to identify this isolate and the latter misidentified the isolate as Pantoea spp. with 98% probability (Table 1). To solve the discrepancy, full-length 16S ribosomal RNA (rRNA) gene sequencing was performed for species-level identification, according to the guidelines of the Clinical and Laboratory Standards Institute (CLSI) [1]. The amplicon was sequenced by Macrogen (Korea) to yield 16S rRNA gene sequence. The length of the sequence was 1455 bp. BLAST (http://www.ncbi.nlm.nih.gov/) search was conducted and the EzBioCloud (http://eztaxon-e.ezbiocloud.net/) [2] was employed using the sequence. For species level, we considered over 99% homology as acceptable criteria. However, we could not identify the sequence accurately because some of the sequences with over 99% homology included the 16S rRNA gene sequence of K. cowanii, Salmonella bongori, and Atlantibacter hermannii. Based on a previous report, we additionally performed gyrB sequencing using the primers UP1 and 181r [3]. Finally, the microorganism was identified as K. cowanii with 99.04% identity, based on similarity with 521 base pairs of K. cowanii strain Pa82 (GenBank accession number CP069319.1).

The antimicrobial susceptibility test (AST) was performed using MicroScan (Beckman Coulter, CA, USA) (Table 2). Antimicrobial susceptibility was interpreted according to the CLSI M100 guidelines [4]. This isolate was resistant to ampicillin, ampicillin/sulbactam, amoxicillin/clavulanic acid, cefoxitin, and cefuroxime. It was susceptible to penicillin in a dose dependent manner, intermediately susceptible to colistin, and highly susceptible to the other tested antibiotics. Due to the concentration range of antibiotics, the minimal inhibitory concentrations (MIC) of ciprofloxacin and levofloxacin could not be interpreted precisely by MicroScan. We also performed Etest (bioMérieux) (Table 2). The Etest results were concordant with the MicroScan results, except in the case of levofloxacin and amoxicillin/clavulanic acid. The levofloxacin MIC in the Etest was interpreted as susceptible. Interestingly, the MIC of amoxicillin/clavulanic acid was significantly different, resulting in a “resistant” result according to MicroScan and a “susceptible” result according to Etest.

The name Enterobacter cowanii was first proposed in 2000 and changed to K. cowanii in 2013 [5, 6]. The Kosakonia spp. was differentiated from the Enterobacter spp. using the multilocus sequence analysis method, which analyzed the genes of gyrB, rpoB, infB, and atpD [6]. There are nine species of Kosakonia, including K. cowanii. K. cowanii belongs to the family Enterobacteriaceae, order Enterobacterales, and class Gammaproteobacteria. It is a gram-negative rod, facultatively anaerobic, oxidase-negative, catalase-positive, and non-spore-forming bacterium, which ferments glucose and reduces nitrates to nitrites [7]. Most K. cowanii individuals are motile and have peritrichous flagella; they grow well on MacConkey agar and have pink, smooth, convex, punctate, umbilicated, glistening colonies [7].

K. cowanii is reported to cause bacterial wilt in some plants [8, 9]; however, its human pathogenicity is not well known. Few case reports have described K. cowanii infection, including bacteremia in neonates [3], rhabdomyolysis as a result of bacteremia preceded by a rose thorn prick [10], and acute cholecystitis [11]. K. cowanii infection seems to occur from exogenous sources [10] and endogenous sources, such as the patient’s microbiota [3, 11]. The possibility of K. cowanii infection must be considered in patients with plant-related injuries. In this case, the infectious synovitis was caused by the prick of a beanstalk.

K. cowanii grows well in routine laboratory cultures; however, its identification method is limited. The VITEK MS (bioMérieux) and VITEK 2 systems (bioMérieux) provide no information on K. cowanii and the VITEK 2 system (bioMérieux) misidentified it as Pantoea spp., which was the same as that obtained in a previous report [3, 11]. Additionally, the colony morphology is very similar between the Pantoea spp., and K. cowanii, making it difficult to differentiate the two organisms [7, 12]. The biochemical characteristics of K. cowanii are similar to those of Pantoea spp., but if the organisms produce gases during glucose, dulcitol or sorbitol fermentation, they are more likely to be K. cowanii [5]. In this case, K. cowanii showed the same biochemical characteristics as previously reported (Table 1) [5, 7], and importantly, with positive sorbitol fermentation. For Pantoea agglomerans ATCC 27155 and Pantoea dispersa ATCC 14589, sorbitol fermentation is negative [13], although there are reports of some Pantoea spp. isolates showing sorbitol fermentation [14, 15]. In addition, full-length 16S rRNA sequencing is insufficient to differentiate among K. cowanii, S. bongori, and A. hermannii; therefore, 16S rRNA with additional gyrB sequencing is the most accurate method to identify K. cowanii. The Bruker Biotyper (Bruker Daltonics) with reliable scores is also conveniently available in clinical laboratories. Additionally, a positive sorbitol fermentation test can indicate the microorganism to be K. cowanii even if the VITEK 2 system (bioMérieux) identifies it as Pantoea spp. in patients who are pricked by plants or have a plant-related job.

The antibiograms of the isolates in this study and previous reports are shown in Table 2 [3, 10, 11]. The Kosakonia genomes were known to lack AmpC β-lactamase [16]; however, in this study, the isolate seemed to have a mechanism for acquired resistance. The isolate was resistant to first- and second-generation cephalosporin antibiotics; therefore, it was likely to be resistant to cefazolin. In general, cefazolin is widely used for skin and soft tissue infections; however, it may not be appropriate in this case. It may be appropriate to use carbapenems, third- or fourth-generation cephalosporins, aminoglycosides, fluoroquinolones, or trimetho-prim/sulfamethoxazole. We also performed some antibiotic susceptibility tests using Etest, most of which showed concordant results with MicroScan. However, the AST results for amoxicillin/clavulanic acid showed discrepancies. Some reports about the discrepancy between the broth microdilution method and other beta-lactams/beta-lactamase inhibitors for Enterobacteriaceae exist [17-19], which may be because beta-lactamase is better expressed in liquid media. Therefore, Etest for K. cowanii should be avoided, especially with beta-lactam/beta-lactamase inhibitors. Meanwhile, Pantoea spp. is also usually susceptible to many antibiotics, but resistance to amoxicillin/clavulanic acid or ampicillin appears to be common [20].

The limitation of this study was that we could not follow-up on the patient’s prognosis. In addition, we did not perform cefazolin AST. Therefore, we could not predict or evaluate the response of this isolate to cefazolin treatment. Finally, whole-genome sequencing was not performed. This might be helpful in determining the antimicrobial resistance mechanisms of K. cowanii.

In conclusion, K. cowanii can cause infectious tenosynovitis, and this is the first case of human infection in Korea. 16S rRNA sequencing is not enough to identify K. cowanii, and gyrB sequencing should be performed as well. Bruker Biotyper (Bruker Daltonics) also seems to be good for identify it. A positive sorbitol fermentation test can indicate the microorganism to be K. cowanii even if the VITEK 2 system (bioMérieux) detects the presence of Pantoea spp. Additionally, K. cowanii may be resistant to beta-lactam antibiotics, including cefazolin; therefore, it may be appropriate to use carbapenems, third- or fourth-generation cephalosporins, aminoglycosides, fluoroquinolones, and trimethoprim/sulfamethoxazole. Furthermore, AST should be conducted using a reference method, and AST using the Etest is unreliable for K. cowanii, especially with beta-lactam/beta-lactamase inhibitors.