Relative Prevalence and Antimicrobial Susceptibility of Clinical Isolates of Elizabethkingia Species Based on 16S rRNA Gene Sequencing

ABSTRACT

Some of the previously reported clinical isolates of Elizabethkingia meningoseptica may be later named species of Elizabethkingia. We determined the accuracy of species identification (with two matrix-assisted laser desorption ionization–time of flight mass spectrometry [MALDI-TOF MS] systems and the Vitek 2 GN card), relative prevalence of three Elizabethkingia spp. in clinical specimens, and antimicrobial susceptibility of the species identified by 16S rRNA gene sequencing. Specimens for culture were collected from patients in a university hospital in Seoul, South Korea, between 2009 and 2015. All 3 Elizabethkingia spp. were detected in patients; among the 86 isolates identified by 16S rRNA gene sequencing, 17 (19.8%) were E. meningoseptica, 18 (20.9%) were Elizabethkingia miricola, and 51 (59.3%) were Elizabethkingia anophelis. Only the MALDI-TOF Vitek MS system with an amended database correctly identified all of the isolates. The majority (76.7%) of the isolates were from the lower respiratory tract, and 8 (9.3%) were from blood. Over 90% of E. meningoseptica and E. anophelis isolates were susceptible to piperacillin-tazobactam and rifampin. In contrast, all E. miricola isolates were susceptible to fluoroquinolones except ciprofloxacin. Further studies are urgently needed to determine the optimal antimicrobial agents for the treatment of infections due to each individual Elizabethkingia species.

INTRODUCTION

Elizabethkingia species are aerobic, nonmotile, oxidase-positive, indole-positive, Gram-negative bacilli that do not ferment glucose. Elizabethkingia spp. can be found frequently in soil, freshwater, salt water, and in hospital environments (1). However, they do not normally exist in the human body. Elizabethkingia meningoseptica (formerly Chryseobacterium meningosepticum) has been a well-known human pathogen since its first description in a case of neonatal meningitis by Elizabeth O. King in 1959 (2). This organism was reported to cause various invasive infections in immunocompromised hosts and to be associated with nosocomial infections and outbreaks in intensive care units (ICUs) (3–5). It has been considered that the incidence of E. meningoseptica bacteremia has increased over the last decade (6). Two new species of Elizabethkingia, Elizabethkingia miricola and Elizabethkingia anophelis, were proposed in 2003 and 2011, respectively (7–9). Therefore, some of the previously reported clinical isolates of E. meningoseptica may be later named species of Elizabethkingia. The first case of E. miricola sepsis was reported in 2008 (10), and a case of E. anophelis neonatal meningitis was reported in 2013 (11).

E. meningoseptica isolates are often resistant (R) to multiple β-lactam antibiotics due to intrinsic class A extended-spectrum β-lactamases (ESBLs) and inherent class B metallo-β-lactamases (MBLs) (12). The antimicrobial susceptibility of Elizabethkingia may vary depending on the species. However, there are scanty data on the susceptibility of the new species. The aims of this study were to determine the accuracy of species identification systems and the relative prevalence of three Elizabethkingia spp. in clinical specimens and to compare the antimicrobial susceptibility of the species identified by 16S rRNA gene sequencing.

RESULTS

Elizabethkingia spp. identified.

Among the total 86 Elizabethkingia isolates, the species identified using 16S rRNA gene sequencing were 17 isolates (19.8%) of E. meningoseptica (99.5% to 99.9% nucleotide identity to E. meningoseptica type strain ATCC 13253), 18 isolates (20.9%) of E. miricola (98.9% to 99.8% nucleotide identity to E. miricola type strain GTC862), and 51 isolates (59.3%) of E. anophelis (99.1% to 100.0% nucleotide identity to E. anophelis type strain FMS-007).

The matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) Vitek MS system with an amended database correctly identified all of the 17, 18, and 51 isolates of E. meningoseptica, E. miricola, and E. anophelis, respectively. However, the Bruker Biotyper correctly identified 16 of 17 E. meningoseptica isolates and 17 of 18 E. miricola isolates but none of the E. anophelis isolates (Table 1). The Vitek 2 GN card system correctly identified 16 of 17 E. meningoseptica isolates but none of the other species.

TABLE 1.

Comparison of species identified by 16S rRNA gene sequencing with those by the two MALDI-TOF systems and the Vitek 2 GN card system

16S rRNA gene sequencing (no. of isolates)	MALDI-TOF Vitek MSa (no. of isolates)	MALDI-TOF Bruker Biotyper (no. of isolates)	Vitek 2 with GN card (no. of isolates)
E. meningoseptica (17)	E. meningoseptica (17)	E. meningoseptica (16)	E. meningoseptica (16)
		Chryseobacterium indologenes (1)	C. indologenes (1)
E. miricola (18)	E. miricola (18)	E. miricola (17)	E. meningoseptica (16)
		C. indologenes (1)	C. indologenes (2)
E. anophelis (51)	E. anophelis (51)	E. meningoseptica (49)	E. meningoseptica (48)
			C. indologenes (1)
		E. meningoseptica/E. miricola (1)	E. meningoseptica (2)
		E. miricola (1)

^aIdentification was based on a SARAMIS database amended with Elizabethkingia spp. spectra provided to bioMérieux.

Among the 86 isolates of Elizabethkingia spp., 66 (76.7%) were recovered from the lower respiratory tract, 8 (9.3%) from blood, 7 (8.1%) from urine, and 5 (5.8%) from other specimens (Table 2). Among the 8 isolates from blood, 2 isolates were identified as E. miricola and 6 isolates were identified as E. anophelis. There were no E. meningoseptica isolates from blood. Careful clinical evaluation suggested that the positive blood cultures in 4 of the 8 patients were not significant. However, they may have been transient invaders from central arterial or venous lines or endotracheal tubes. Therefore, no antimicrobial agents were administered for Elizabethkingia infection (Table 3). All of the 8 patients had various underlying diseases and had indwelling catheters or endotracheal tubes. Four patients were admitted to the ICU. All of the patients, except for one, were treated with various antimicrobial agents during the 7 days prior to blood culture. Among the four patients, two were cured with tigecycline and trimethoprim-sulfamethoxazole, to which the isolates were susceptible (S). The remaining two patients were cured with trimethoprim-sulfamethoxazole, although the MICs for the isolates were 4 and 76 μg/ml (resistance breakpoints, ≥4 and 76 μg/ml), respectively.

TABLE 2.

Source of detection of 86 isolates of Elizabethkingia spp. at a tertiary care hospital from 2009 to 2015

Species (no. of isolates)	No. (%) of isolates from:
	Lower respiratory	Blood	Urine	Othera
E. meningoseptica (17)	14	0	3	0
E. miricola (18)	12	2	0	4
E. anophelis (51)	40	6	4	1
Total (%)	66 (76.7)	8 (9.3)	7 (8.1)	5 (5.8)

^aOther sources includes eye, neck, head, pleural fluid, and external ear fluid.

TABLE 3.

Case summary of eight patients with Elizabethkingia spp. isolation from blood

No.	Age/sexa	Initial identificationb	Final identificationc	Underlying diseased	Indwelling device	Warde	No. positive/total pairs, bottles (culture date)	Infection signf		Previous antibiotic therapy (1 week before culture)	Antimicrobial therapy for Elizabethkingia infection	Outcome
								BT (°C)	WBC (per μl)			Microbiologic eradicationg	Clinical responseh
1	79/M	E. meningoseptica	E. miricola	Bladder cancer, hypoxic brain damage	Endotracheal tube, central venous line	ICU	1/3 (Aug. 8, 2011)	38.0	23,990	None	None	NAi	NA
							1/1 (Aug. 9, 2011)
2	69/M	E. meningoseptica	E. miricola	COPD, CRF, cerebral infarction, fungal pneumonia	Endotracheal tube	GW	1/4 (Aug. 26, 2011)	35.9	9,990	Meropenem, teicoplanin, levofloxacin	None	NA	NA
3	69/M	E. meningoseptica	E. anophelis	Rectal cancer	Endotracheal tube, arterial line, central venous line	ICU	1/3 (Sep. 25, 2010)	38.0	8,320	Imipenem, metronidazole	None	NA	NA
4	72/F	E. meningoseptica	E. anophelis	Klatskin tumor, DM	Endotracheal tube, arterial line, central venous line	GW	2/3 (Oct. 30, 2010)	36.2	7,420	Tigecycline	Tigecycline, ciprofloxacin	Cured	Failed
5	72/M	E. meningoseptica	E. anophelis	Alcoholic LC	Endotracheal tube, central venous line	GW	3/3 (Nov. 13, 2012)	37.8	2,400	Imipenem, colistin, teicoplanin, minocycline	Trimethoprim-sulfamethoxazole	Cured	Cured
							2/2 (Nov. 15, 2012)
6	67/F	E. meningoseptica	E. anophelis	Myelofibrosis, splenomegaly, thrombotic endocarditis	Central venous line	GW	4/4 (Nov. 22, 2012)	37.5	11,180	Ciprofloxacin, cefazolin	Trimethoprim-sulfamethoxazole	Cured	Cured
							1/3 (Nov. 23, 2012
							1/1 (Nov. 24, 2012)
							2/4 (Nov. 26, 2012)
7	49/M	E. meningoseptica	E. anophelis	Mitral valve replacement, valvular heart failure	Endotracheal tube, arterial line	ICU	1/5 (Mar. 20, 2013)	38.6	16,630	Piperacillin-tazobactam, cefepime, teicoplanin	None	NA	NA
8	63/M	E. meningoseptica	E. anophelis	Pneumonectomy, lung transplantation, RA	Endotracheal tube, chest tube, arterial line, central venous line	ICU	2/3 (Sep. 20, 2013)	36.4	2,890	Meropenem, colistin	Trimethoprim-sulfamethoxazole, tigecycline	Cured	Cured
							2/3 (Sep. 22, 2013)
							4/4 (Sep. 24, 2013)

^aM, male; F, female.

^bIdentified by Vitek 2 with GN card for no. 1 to 7 and by MALDI-TOF Bruker Biotyper for No. 8.

^cFinal identification was done by 16S rRNA gene sequencing.

^dCOPD, chronic obstructive pulmonary disease; CRF, chronic renal failure; LC, liver cirrhosis; RA, rheumatic arthritis; DM, diabetes mellitus.

^eGW, general ward.

^fBT, body temperature; WBC, white blood cell.

^gMicrobiologic eradication was the absence of the original pathogens detected from blood (7 days after the first positive blood culture).

^hClinical response: a favorable clinical response was defined as the resolution of fever (defined as ≥38.0°C), leukocytosis (WBC, ≥11 × 10⁶/μl), and hypotension (mean arterial pressure of <65 mm Hg), in addition to no longer requiring support from vasoactive agents. Patients who had persistence or deterioration in clinical parameters or who died were classified as treatment failures.

ⁱNA, not applicable. There was no follow up data due to transfer of the patient within 7 days.

In pulsed-field gel electrophoresis (PFGE) analysis, isolates of each Elizabethkingia species belonged to 5 to 10 different PFGE groups, while identical pulsotypes were found in 8 of 17 E. meningoseptica isolates, 6 of 18 E. miricola isolates, and 17 of 51 E. anophelis isolates (see Fig. S1 in the supplemental material).

Antimicrobial susceptibilities.

The MICs of the antimicrobial agents and the susceptibilities of the isolates are shown in Table 4 (see Data Set S1 in the supplemental material). Among the E. meningoseptica isolates, 100% and 94% were susceptible to piperacillin-tazobactam and rifampin, respectively, but only 23% to 41% were susceptible to fluoroquinolones. Unlike E. meningoseptica, all E. miricola isolates were susceptible to fluoroquinolones, except for ciprofloxacin. Over 90% of E. meningoseptica and E. anophelis isolates were susceptible to piperacillin-tazobactam and rifampin. Although none of the species were susceptible to vancomycin, all three species exhibited at least 94% intermediate (I) reaction to this agent.

TABLE 4.

Antimicrobial susceptibilities of Elizabethkingia isolates determined by the agar dilution method

Species (no. of isolates) and antimicrobial agents	Breakpoint (μg/ml)a		MIC (μg/ml)			Susceptibility (%)
	S	R	Range	50%	90%	S	I	R
E. meningoseptica (17)
Piperacillin	≤16	≥128	16–32	16	32	65	35	0
Piperacillin-tazobactamb	≤16	≥128	8–16	8	16	100	0	0
Ceftazidime	≤8	≥32	64 to >128	>128	>128	0	0	100
Imipenem	≤4	≥16	16–32	32	32	0	0	100
Ciprofloxacin	≤1	≥4	1 to >64	64	>64	23	6	71
Levofloxacin	≤2	≥8	0.5–128	16	64	35	0	65
Moxifloxacin	≤2	≥8	0.12–64	4	32	41	12	47
Gatifloxacin	≤2	≥8	0.5–128	8	64	35	12	53
Trimethoprim-sulfamethoxazoleb	≤2	≥4	2–8	4	4	6	0	94
Gentamicin	≤4	≥16	4 to >128	32	64	6	0	94
Vancomycin	≤4	≥32	8–64	8	16	0	94	6
Rifampin	≤1	≥4	0.25–2	0.5	1	94	6	0
E. miricola (18)
Piperacillin	≤16	≥128	4–32	16	32	83	17	0
Piperacillin-tazobactam	≤16	≥128	4–32	8	16	94	6	0
Ceftazidime	≤8	≥32	64 to ≥128	>128	>128	0	0	100
Imipenem	≤4	≥16	16 to ≥64	64	64	0	0	100
Ciprofloxacin	≤1	≥4	0.5–4	1	4	56	22	22
Levofloxacin	≤2	≥8	0.25–2	0.5	2	100	0	0
Moxifloxacin	≤2	≥8	≤0.06–1	0.25	1	100	0	0
Gatifloxacin	≤2	≥8	0.12–2	0.5	2	100	0	0
Trimethoprim-sulfamethoxazole	≤2	≥4	1–8	4	8	28	0	72
Gentamicin	≤4	≥16	4 to >128	8	>128	45	22	33
Vancomycin	≤4	≥32	8–16	16	16	0	100	0
Rifampin	≤1	≥4	0.25 to >128	1	16	66	17	17
E. anophelis (51)
Piperacillin	≤16	≥128	8–64	16	32	82	18	0
Piperacillin-tazobactam	≤16	≥128	≤0.12–32	8	8	92	8	0
Ceftazidime	≤8	≥32	64 to >128	>128	>128	0	0	100
Imipenem	≤4	≥16	16 to >64	64	>64	0	0	100
Ciprofloxacin	≤1	≥4	1 to >64	64	>64	22	6	72
Levofloxacin	≤2	≥8	0.5 to >128	32	64	29	6	65
Moxifloxacin	≤2	≥8	0.12–64	4	32	41	10	49
Gatifloxacin	≤2	≥8	0.25–128	8	32	33	12	55
Trimethoprim-sulfamethoxazole	≤2	≥4	2–16	4	8	22	0	78
Gentamicin	≤4	≥16	1 to >128	32	64	22	23	55
Vancomycin	≤4	≥32	8–64	16	16	0	94	6
Rifampin	≤1	≥4	≤0.06–16	1	1	96	2	2

^aThe interpretive criteria applied were those of the CLSI for non-Enterobacteriaceae; the criteria for vancomycin and rifampin were those for Staphylococcus or Enterococcus spp. The criterion of gatifloxacin was that for moxifloxacin.

^bIn the combinations, the concentration of tazobactam was 4 μg/ml constant, and the ratio of trimethoprim to sulfamethoxazole was 1 to 19.

DISCUSSION

The accuracy of species identification was low with the Vitek 2 GN card system. Although our Bruker Biotyper without an amended database failed to identify E. anophelis isolates, the addition of a database for E. anophelis was recently reported (13). This result indicates that a MALDI-TOF mass spectrometry (MS) system can be a reliable species identification system for the genus Elizabethkingia.

Although E. meningoseptica infections are well known, E. miricola sepsis in a lymphoma patient was reported in the United States after the proposal of new species (10). Several E. anophelis infections have been reported from tropical or subtropical regions, the Central African Republic (11), Singapore (14), and Hong Kong (15). A recent study in Hong Kong urged researchers to consider the clinically significant morbidity and mortality of patients with E. anophelis bacteremia (13). An E. anophelis outbreak in Wisconsin in 2016 resulted in the deaths of at least 18 patients (16).

In our study, which took place in the temperate country of South Korea, all three species of Elizabethkingia were detected in patients, and E. anophelis was the most prevalent. It is interesting that E. meningoseptica was not present while E. anophelis was the most common among blood isolates (Table 2). These findings suggest that E. anophelis plays a significant role as a human pathogen. In general, the majority of blood isolates are clinically significant (5, 13, 16). However, in our study, only 4 of the 8 patients with positive Elizabethkingia blood cultures were clinically significant, although all of the patients had risk factors for infection (Table 3). The majority of the Elizabethkingia isolates were detected in lower respiratory tract specimens, but it was difficult to distinguish infection from colonization as reported in other studies (4, 17).

Our PFGE analysis of each species showed that certain pulsotypes were more prevalent than others, suggesting that these types are either more prevalent in the hospital environment or that they have a higher capability to infect or colonize.

E. meningoseptica has been known to be resistant to multiple antimicrobial agents (18, 19). However, as mentioned above, the susceptibility of E. meningoseptica in the previous study may include those of the other 2 species. To the best of our knowledge, our study is the first one to compare the susceptibilities of all Elizabethkingia species (Table 4). The organisms are typically resistant to β-lactams (15, 18). In a previous study from our group (19), all 31 isolates of E. meningoseptica (which may include other Elizabethkingia species) had both bla_BlaB and bla_GOB genes. The GenBank database shows that chromosomes of E. anophelis (accession numbers CP006576 and CP007547) and E. miricola (accession number CP011059) possess bla_BlaB, bla_GOB, and bla_CME genes and an AmpC β-lactamase gene. In our study, over 90% of the isolates of 3 Elizabethkingia spp. were susceptible to a piperacillin-tazobactam combination. This has also been shown by other studies (5, 20). However, it is necessary to evaluate clinical efficacy, given that all Elizabethkingia spp. appear to be inherent MBL producers.

Antimicrobial resistance may vary depending on the species as well as the region and time of bacterial isolation. As mentioned above, limited data are currently available on the susceptibility patterns of Elizabethkingia spp. Although three previous studies were performed using relatively significant numbers of isolates, the specimens were from unspecified sources of patients, blood, or hospital environments (5, 21, 22). Furthermore, all three reports stated that the species were E. meningoseptica, but the isolates were identified before the proposal of new species or unreliable phenotypic methods were used for identification. However, by comparing our results, based on identification by 16S rRNA sequencing, to those of other studies, the following generalization can be made: Elizabethkingia spp. are nonsusceptible to ceftazidime, imipenem, and vancomycin (high vancomycin susceptibility in a study may be due to the use of a higher breakpoint, 16 μg/ml [21]); the susceptibility rates of E. miricola to fluoroquinolones are higher than those of the other species; and the susceptibility of Elizabethkingia spp. to other antimicrobial agents are difficult to predict.

Several reports have shown that the incidence of Elizabethkingia bacteremia increased and the mortality rate was high (5, 20, 23, 24). Indwelling devices and inappropriate antimicrobial therapy were independent risk factors for poor outcomes with Elizabethkingia bacteremia (5, 24, 25). In our study, all 4 bacteremic patients were microbiologically cured with trimethoprim-sulfamethoxazole alone or with a combination of tigecycline plus trimethoprim-sulfamethoxazole or ciprofloxacin. Anecdotal reports have indicated that some cases of E. meningoseptica infection respond only to combinations of piperacillin-tazobactam plus rifampin, vancomycin plus rifampin, or a fluoroquinolone plus vancomycin and rifampin (6). In our study, none of the Elizabethkingia isolates were susceptible to vancomycin, and the majority were intermediate, indicating similar susceptibility with those of the worldwide collection from 1999 to 2001 (22). Therefore, it seems that vancomycin alone is ineffective in the treatment of Elizabethkingia infection.

In conclusion, E. anophelis was the most frequently detected species in clinical specimens. Over 90% of 3 Elizabethkingia spp. were susceptible to piperacillin-tazobactam. The majority of E. meningoseptica and E. anophelis isolates were susceptible to rifampin, and all isolates of E. miricola only were susceptible to levofloxacin, moxifloxacin, and gatifloxacin. Therefore, further studies are urgently needed to determine the optimal antimicrobial agents for treatment of infections caused by each individual Elizabethkingia species.

MATERIALS AND METHODS

Clinical specimens and identification of Elizabethkingia spp.

Clinical specimens for bacterial culture were collected from patients at a tertiary care university hospital in Seoul, South Korea between January 2009 and February 2015. The species were initially identified using the Vitek 2 GN card system (bioMérieux, Mercy l'Etoile, France). Isolates identified as either Elizabethkingia spp. or Chryseobacterium spp. were kept frozen until used in this study.

16S rRNA gene sequencing and MALDI-TOF MS analysis.

The 16S rRNA gene was amplified and sequenced using the universal primers 8F (5′-AGA GTT TGA TCC TGG CTC AG-3′) and 1541R (5′-AAG GAG GTG ATC CAG CCG CA-3′). The following additional primers were used to analyze the sequence: 310R (5′-AGT ACC AGT GTG GGG GAT CA-3′) and 1170F (5′-CAA ATC ATC ACG GCC CTT AC-3′). The species were identified by comparing the sequences using the EzTaxon server (http://www.ezbiocloud.net/).

All clinical isolates were identified by two MALDI-TOF systems, the Bruker Biotyper (Bruker Daltonics, Bremen, Germany) and the Vitek MS (bioMérieux). There were no reference data for the identification of E. anophelis in either system. However, the Vitek MS research use only (RUO) (Saramis) database was amended for our study by providing the spectra data of 20 isolates of three Elizabethkingia spp. identified by 16S rRNA gene sequencing to the bioMérieux. These were used to compute species-specific SuperSpectra for automated identification with SARAMIS (details to be published elsewhere). The accuracy of species identification using the MALDI-TOF and Vitek 2 GN card systems was determined by comparing the results of the 16S rRNA gene sequence as a reference.

Pulsed-field gel electrophoresis.

Chromosomal DNA of Elizabethkingia isolates were digested with XbaI and analyzed for PFGE patterns using the CHEF DR II system (Bio-Rad, Hercules, CA, USA) as described previously (15).

Antimicrobial susceptibility testing.

The MICs of the antimicrobial agents were determined using an agar dilution method (26). The antimicrobial agents used were piperacillin and tazobactam (Wyeth, Pearl River, NY, USA); ceftazidime, gentamicin, rifampin, and vancomycin (Sigma Chemical, St. Louis, MO, USA); imipenem (Choongwae, Seoul, South Korea); ciprofloxacin and moxifloxacin (Bayer Korea, Seoul, South Korea); levofloxacin (Daiichi, Tokyo, Japan); gatifloxacin (Bristol-Myers Squibb, Princeton, NJ, USA); and trimethoprim and sulfamethoxazole (Dong Wha, Seoul, South Korea).

The MICs were interpreted based on the Clinical and Laboratory Standards Institute (CLSI) criteria for other non-Enterobacteriaceae (27). The breakpoints used for vancomycin (S, ≤4 μg/ml; R, ≥32 μg/ml) and rifampin (S, ≤1 μg/ml; R, ≥4 μg/ml) were those for Staphylococcus spp. The moxifloxacin breakpoint was used for gatifloxacin. Escherichia coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853, and Staphylococcus aureus ATCC 29213 were used as controls.

Accession number(s).

The GenBank accession numbers of 16S rRNA sequence are as follows: KP836318 and KP836320 for E. meningoseptica; KP836321 and KP844567 for E. miricola; and KT768343, KT768344, KT768345, KP836317, and KP836319 for E. anophelis.