Antimicrobial Susceptibility of Elizabethkingia Species: Report from a Reference Laboratory

ABSTRACT

Elizabethkingia species are Gram-negative bacilli that were most recently linked to a cluster of infections in the Midwestern United States from 2016 to 2017. Inappropriate empirical and directed antibiotic selection for this organism is common among providers and is an independent risk factor for mortality. Trends in antimicrobial susceptibility profiles of Elizabethkingia species from a referral laboratory over a 10-year period were reviewed. Identification methods used over time varied and included biochemical panels, matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS), and 16S rRNA gene sequencing. Agar dilution was used to conduct antimicrobial susceptibility testing. One hundred seventy-four clinical isolates were included. The lower respiratory tract (20/37; 54%) was the most common specimen source in pediatric patients, whereas blood isolates (62/137; 45%) constituted the most prevalent source in adults. Among the identified species, Elizabethkingia meningoseptica (72/121; 59%) constituted the majority. All Elizabethkingia species tested against minocycline were susceptible (18/18; 100%), and 90% of isolates tested against trimethoprim-sulfamethoxazole (TMP-SMX) (117/130) were susceptible. Of the 12 Elizabethkingia miricola isolates, most of the tested isolates were susceptible to piperacillin-tazobactam (11/12; 92%) and levofloxacin (11/12; 92%), whereas the Elizabethkingia anophelis isolates most often tested susceptible to piperacillin-tazobactam (13/14; 93%). In this study, Elizabethkingia species showed high rates of in vitro susceptibility to minocycline and TMP-SMX. Further studies are needed to investigate the clinical implications of species-level differences in antimicrobial susceptibilities in this genus.

INTRODUCTION

Elizabethkingia species are aerobic, oxidase-positive, glucose-nonfermenting, nonmotile Gram-negative bacilli found in water and soil (1). Although rare pathogens, Elizabethkingia species can cause life-threatening infections and have been implicated in human infections since their description by Elizabeth O. King in 1959 when they were first reported as a cause of an outbreak of meningitis in infants (1, 2). A novel species, Elizabethkingia anophelis, was described in 2011 and linked to sporadic hospital outbreaks in the United States (3–6), Taiwan (7), and Hong Kong (2). In 2016, the U.S. Centers for Disease Control and Prevention (CDC) investigated a cluster of E. anophelis infections in Wisconsin, Michigan, and Illinois, which were related to sporadic infections rather than a single-point-source outbreak (4, 6).

Elizabethkingia species are resistant to most β-lactams, colistin, and aminoglycosides and are variably susceptible to piperacillin-tazobactam, fluoroquinolones, and trimethoprim-sulfamethoxazole (TMP-SMX) (1, 5). Several point mutations in DNA gyrase subunit A (GyrA) and subunit B (GyrB) have been described in association with various degrees of fluoroquinolone resistance in Elizabethkingia species (1). Two chromosomal metallo-β-lactamase genes, bla_BlaB and bla_GOB, confer intrinsic resistance to carbapenems, while the extended-spectrum serine β-lactamase CME (Ambler class D) is associated with cephalosporin resistance in Elizabethkingia meningoseptica (8).

Elizabethkingia infections are associated with high morbidity and mortality rates. In one study, the 28-day mortality rate was 24% (2). Risk factors for a poor prognosis include indwelling vascular catheters, mechanical ventilation, malignancy, diabetes mellitus, renal insufficiency, and liver cirrhosis (9–11). Common infections associated with Elizabethkingia species include pneumonia, urinary tract infections, meningitis, bacteremia, skin and soft tissue infections, and catheter-associated bloodstream infections (1, 9, 10). Appropriate therapeutic choices are crucial for management since inappropriate antimicrobial selection may be associated with increased mortality independent of other risk factors (9, 12). Some laboratories may be unable to test for the antimicrobial susceptibility of these organisms due to limitations of testing platforms and may need to rely on antibiogram patterns or published literature to guide therapy. Therefore, we aimed to retrospectively review in vitro antimicrobial susceptibility testing (AST) profiles of Elizabethkingia species from 2011 through 2021 as assessed at a referral laboratory in the Midwestern United States.

MATERIALS AND METHODS

Study design.

The clinical microbiology laboratory database was reviewed for clinical isolates of Elizabethkingia species from November 2011 to March 2021. Specimens submitted from the local patient population were included, as were isolates sent as referrals for identification and/or antimicrobial susceptibility testing. Demographic and microbiological variables were recorded, including patient age, gender, source and year of specimen collection, and antimicrobial susceptibility results. Duplicate isolates, defined as isolates from the same patient, were excluded regardless of the specimen source.

Microbial identification.

Species-level identification was achieved, when possible, using methods that changed over time to reflect updates in microbiological practice. Before 2014, organisms were identified using the BD Phoenix NID panel (BD, Sparks, MD) in addition to conventional biochemical techniques as necessary. Bruker matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS; Bruker Daltonics) RUO (research use only) and Mayo Clinic-developed libraries were used as the primary methods of identification from 2014 onward. MALDI-TOF MS testing was performed using on-plate extraction as the primary method, with full-tube extraction utilized in instances where identification was not obtained via on-plate extraction. During 2017 to 2019, Elizabethkingia isolates were referred to the CDC for species-level identification. Starting from mid-2019, isolates were routinely reported only to the genus level using MALDI-TOF MS. 16S rRNA gene PCR/sequencing was used for species identification for sterile-source isolates or upon clinician request. On 24 August 2020, due to RUO library updates (Bruker MALDI Biotyper (MBT) compass library DB-8468 Main Spectra Profile (MSP), April 2019), routine reporting of E. meningoseptica, E. anophelis, and E. miricola as the E. meningoseptica group commenced.

Antimicrobial susceptibility testing.

The MIC values for antimicrobial agents were determined using agar dilution according to Clinical and Laboratory Standards Institute (CLSI) methods (13). CLSI breakpoints for “other non-Enterobacterales” were applied per CLSI document M100 guidelines according to annual M100 guidance appropriate for the calendar year (13). Nonetheless, there were no revised breakpoints for the “other non-Enterobacterales” category spanning the years of the study. However, in 2017, the CLSI breakpoints for colistin (≤2 μg/mL for susceptible, 4 μg/mL for intermediate, and ≥8 μg/mL for resistant) were removed from the M100 document, and testing of colistin for this group of isolates was no longer performed. Susceptibility testing was performed on all sterile-site isolates and upon clinician request.

Patient consent statement.

This study was approved by the Mayo Clinic Institutional Review Board (IRB) (approval number 21-004632).

RESULTS

Patient population.

From November 2011 through March 2021, there were 235 clinical isolates of Elizabethkingia species from 174 patients. Following the exclusion of duplicates from patients, 174 isolates were included in this study.

Of the 174 patients, 54% were male, and the mean age was 48 years (standard deviation [SD], ±28.9 years). The patient cohort included 37 (21%) children aged 18 years or younger and 68 (39%) adults older than 65 years. The geographic distribution of institutions from which the isolates originated is represented in Fig. 1. Of 174 isolates, 24 (14%) originated from Minnesota, 23 (13%) were from Illinois, 15 (9%) were from Georgia, and 12 (7%) were from Wisconsin, with the remaining isolates originating from 20 different states.

FIG 1.

Geographic distribution of states of origin for Elizabethkingia species isolates submitted for identification and susceptibility testing from November 2011 to March 2021.

Clinical isolates and identification.

The annual frequency of Elizabethkingia identification increased from 6 in 2012 to 36 in 2020 (Fig. 2). The most common specimens from which organisms were isolated included blood (65/174; 37%), the lower respiratory tract (60/174; 34%), the upper respiratory tract (19/174; 11%), and urine (16/174; 9%). The remaining isolates, which constituted 8% of all specimens, were recovered from skin and soft tissues, corneas, contact lens solutions, and catheter tips. Differences in specimen sources were noted between adults and children, with blood being the most common specimen in adults (62/137; 45%) and the lower respiratory tract being the most common in children (20/37; 54%).

FIG 2.

Number of Elizabethkingia isolates submitted per year for identification and susceptibility testing from January 2012 to December 2020.

Overall, 121 (69%) isolates were identified to the group level (i.e., Elizabethkingia meningoseptica group) or species level. Fifty-three (30%) isolates were identified to the genus level as Elizabethkingia sp. Among isolates identified to the species level, the most common was E. meningoseptica (72/121; 59%), followed by E. anophelis (18/121; 15%) and E. miricola (13/121; 11%). Eighteen isolates were identified as belonging to the E. meningoseptica group (15%), and all were from 2020 and 2021.

Among 65 blood isolates, E. meningoseptica was the most common identification reported (30/65; 46%), followed by Elizabethkingia sp. (13/65; 20%), E. anophelis (11/65; 17%), and E. miricola (4/65; 6%). There were seven blood isolates identified as E. meningoseptica group isolates.

Antimicrobial susceptibility testing.

Antimicrobial susceptibility testing results are summarized in Table 1. Testing of susceptibility to at least one antimicrobial agent was performed on 146 isolates (84%). All isolates that underwent in vitro susceptibility testing for minocycline (n = 18) were susceptible. One hundred seventeen of 130 isolates (90%) tested were susceptible to TMP-SMX, 110/131 (84%) were susceptible to piperacillin-tazobactam, and 105/130 (80%) were susceptible to levofloxacin (Table 1). At least 98% of isolates tested were resistant to colistin, aztreonam, tobramycin, and ceftazidime. Of the clinical isolates tested for antimicrobial susceptibility, E. meningoseptica had high rates of susceptibility to TMP-SMX (42/47; 89%) and piperacillin-tazobactam (42/47; 89%), E. anophelis had a high rate of susceptibility to piperacillin-tazobactam (13/14; 93%), and E. miricola had high rates of susceptibility to levofloxacin (11/12; 92%) and piperacillin-tazobactam (11/12; 92%) (Table 1).

TABLE 1.

Antimicrobial susceptibility test results of Elizabethkingia species^a

Antimicrobial agent	No. of isolates with result/total no. of isolates tested (%)
	E. anophelis			E. meningoseptica			E. miricola			E. meningoseptica group			Elizabethkingia genusb			Total isolates
	S	I	R	S	I	R	S	I	R	S	I	R	S	I	R	S	I	R
Amikacin	0	2/13 (15.4)	11/13 (84.6)	0	6/45 (13.3)	39/45 (86.7)	1/12 (8.3)	0	11/12 (91.7)	0	3/13 (23.1)	10/13 (76.9)	1/44 (2.3)	2/44 (4.5)	41/44 (93.2)	2/127 (1.6)	13/127 (10.2)	112/127 (88.2)
Aztreonam	0	0	13/13 (100)	0	0	30/30 (100)	0	0	7/7 (100)	0	1/12 (8.3)	11/12 (91.7)	0	0	39/39 (100)	0	1/101 (1)	100/101 (99)
Cefepime	0	1/13 (7.7)	12/13 (92.3)	0	3/47 (6.4)	44/47 (93.6)	1/12 (8.3)	0	11/12 (91.7)	0	0	13/13 (100)	0	1/45 (2.2)	44/45 (97.8)	1/130 (0.8)	5/130 (3.8)	124/130 (95.4)
Ceftazidime	0	0	13/13 (100)	0	1/46 (2.2)	45/46 (97.8)	1/12 (8.3)	0	11/12 (91.7)	0	0	12/12 (100)	0	0	44/44 (100)	1/127 (0.8)	1/127 (0.8)	125/127 (98.4)
Ciprofloxacin	5/14 (35.7)	4/14 (28.6)	5/14 (35.7)	11/48 (22.9)	21/48 (43.8)	16/48 (33.3)	6/12 (50)	5/12 (41.7)	1/12 (8.3)	5/13 (38.5)	4/13 (30.8)	4/13 (30.8)	15/44 (34.1)	17/44 (38.6)	12/44 (27.3)	42/131 (32.1)	51/131 (38.9)	38/131 (29)
Colistin	0	0	1/1 (100)	0	0	26/26 (100)	0	0	4/4 (100)	0	0	0	0	0	12/12 (100)	0	0	43/43 (100)
Doxycycline	0	0	0	2/3 (66.7)	1/3 (33.3)	0	0	0	0	0	0	0	1/1 (100)	0	0	3/4 (75)	1/4 (25)	0
Gentamicin	0	2/13 (15.4)	11/13 (84.6)	1/45 (2.2)	5/45 (11.1)	39/45 (86.7)	1/12 (8.3)	5/12 (41.7)	6/12 (50)	3/13 (23.1)	0	10/13 (76.9)	4/45 (8.9)	6/45 (13.3)	35/45 (77.8)	9/128 (7)	18/128 (14.1)	101/128 (78.9)
Levofloxacin	10/14 (71.4)	1/14 (7.1)	3/14 (21.4)	37/47 (78.7)	3/47 (6.4)	7/47 (14.9)	11/12 (91.7)	1/12 (8.3)	0	11/12 (91.7)	1/12 (8.3)	0	36/45 (80)	4/45 (8.9)	5/45 (11.1)	105/130 (80.8)	10/130 (7.7)	15/130 (11.5)
Meropenem	0	0	13/13 (100)	1/46 (2.2)	0	45/46 (97.8)	1/12 (8.3)	0	11/12 (91.7)	0	0	13/13 (100)	0	0	45/45 (100)	2/129 (1.6)	0	127/129 (98.4)
Minocycline	2/2 (100)	0	0	11/11 (100)	0	0	0	0	0	1/1 (100)	0	0	4/4 (100)	0	0	18/18 (100)	0	0
Piperacillin-tazobactam	13/14 (92.9)	0	1/14 (7.1)	42/47 (89.4)	5/47 (10.6)	0	11/12 (91.7)	1/12 (8.3)	0	10/13 (76.9)	3/13 (23.1)	0	34/45 (75.6)	7/45 (15.6)	4/45 (8.9)	110/131 (84)	16/131 (12.2)	5/131 (3.8)
TMP-SMX	12/14 (85.7)	0	2/14 (14.3)	42/47 (89.4)	0	5/47 (10.6)	10/12 (83.3)	0	2/12 (16.7)	12/13 (92.3)	0	1/13 (7.7)	41/44 (93.2)	0	3/44 (6.8)	117/130 (90)	0	13/130 (10)
Tobramycin	0	0	13/13 (100)	0	0	45/45 (100)	0	0	12/12 (100)	0	0	13/13 (100)	0	0	44/44 (100)	0	0	127/127 (100)

^aS, susceptible; I, intermediate; R, resistant; TMP-SMX, trimethoprim-sulfamethoxazole.

^bElizabethkingia isolates identified only to the genus level.

Of 65 blood isolates, testing of susceptibility to at least one antimicrobial agent was performed for 55 isolates (85%). All tested blood isolates were susceptible to minocycline (n = 12) but resistant to colistin (n = 15), tobramycin (n = 46), and aztreonam (n = 46). Furthermore, 87% of blood isolates tested were susceptible to piperacillin-tazobactam, 85% were susceptible to TMP-SMX, and 79% were susceptible to levofloxacin. Rates of susceptibility were low for blood isolates for amikacin (4%), ceftazidime (2%), ciprofloxacin (33%), cefepime (2%), gentamicin (6%), and meropenem (4%).

DISCUSSION

Infections due to Elizabethkingia lead to high morbidity and mortality rates, in part because patients are often critically ill but also likely because optimal therapy is poorly defined (1, 2). Compared to bloodstream infections caused by other glucose-nonfermenting Gram-negative bacilli, E. meningoseptica bacteremia is associated with a low rate of effective empirical antibiotic therapy within 48 h of blood culture collection and a long time to targeted antibiotic administration (14). Chen et al. reported on 70 critically ill patients with E. meningoseptica bacteremia; those who received a cephalosporin after susceptibility test results were available had an 11% survival rate, whereas 75% of patients who received piperacillin-tazobactam survived (14). Similarly, Lin et al. reported a case series of 67 patients with E. anophelis infection (9). Seventy-five percent of patients received ineffective empirical antimicrobial therapy, which was an independent risk factor for mortality (9).

In previous in vitro susceptibility studies using CLSI breakpoints (9, 15, 16), Elizabethkingia species were frequently resistant to antimicrobials commonly used in clinical practice, such as β-lactams (including broad-spectrum cephalosporins and carbapenems), colistin, and aminoglycosides. Rifampin (16) and minocycline (9, 17–19) showed favorable in vitro activity against some Elizabethkingia species. In one study, time-kill assays demonstrated potent activity of minocycline against E. anophelis biofilm-embedded cells; in addition, MICs by agar dilution for all 30 E. anophelis isolates were ≤2 μg/mL (17). In our study, all 18 isolates tested for minocycline demonstrated MICs in the susceptible range. Although there were only a limited number of isolates tested against minocycline, these findings support the high susceptibility rate demonstrated in previous studies. The use of vancomycin as monotherapy is clinically ineffective for the treatment of Elizabethkingia infections, which is consistent with elevated MICs of ≥4 μg/mL (1, 20). Given these issues and the lack of supportive microbiological testing data from breakpoint-setting organizations, many laboratories, including ours, do not assess vancomycin susceptibility in Elizabethkingia species.

Previous studies reported worldwide geographic variability in Elizabethkingia rates of susceptibility to fluoroquinolones, piperacillin-tazobactam, and TMP-SMX (1, 9, 15, 17). For instance, Han et al. reported 86 clinical isolates from South Korea with TMP-SMX susceptibility rates of 6 to 28% among different species, assessed using agar dilution (Table 2) (21). Two other reports from Taiwan showed different susceptibility rates (0 to 18% versus 97 to 100%) using broth microdilution methods; however, susceptibility testing platforms were different between these two studies (18, 19). All three studies applied CLSI breakpoints (18, 19, 21). In contrast to these reports (19, 21), which reported in vitro rates of susceptibility to TMP-SMX of <30%, the TMP-SMX susceptibility rate was 90% in our cohort, which is similar to the findings for the outbreak isolates from Wisconsin between 2015 and 2016 (5). The differences in susceptibility results across studies could be due to the geographic variability of resistance characteristics. In previous studies, the ciprofloxacin susceptibility rate was between 0 and 56%, whereas the levofloxacin susceptibility rate varied between 16 and 100% against different Elizabethkingia species (18, 19, 21, 22). In the current study, Elizabethkingia MICs were within the susceptible range in 32% (42/131) of all isolates against ciprofloxacin. E. meningoseptica (80%) and E. miricola (92%) had high rates of susceptibility to levofloxacin.

TABLE 2.

Overview of antimicrobial susceptibility testing results for the current study and previous similar studies^a

Parameter	Value for study
	This study	Han et al. (21)	Cheng et al. (19)	Kuo et al. (18)	Wang et al. (22)
Country	USA	South Korea	Taiwan	Taiwan	China
Mean age (yrs) ± SD	48 ± 28.9	DNC	DNC	DNC	64 ± 21
No. (%) of female patients	80 (46)	DNC	DNC	DNC	16 (30.8)
Yrs when isolates were collected	2011–2021	2009–2015	DNC	2002–2018	2012–2015 and 2017–2018
No. of isolates	174	86	269	108	52
Identification test(s)	MALDI-TOF MS (Bruker Biotyper RUO database with Mayo Clinic database)	Vitek 2 with GN card, MALDI-TOF MS Bruker Biotyper and Vitek MS, 16S rRNA gene sequencing	MALDI-TOF MS with updated database	16S rRNA gene sequencing	16S rRNA gene sequencing
No. (%) of isolates from site
Respiratory	79 (45)	66 (77)	DNC	52 (48)	45 (86)
Blood	65 (37)	8 (9)	DNC	52 (48)	2 (4)
Urine	16 (9)	7 (8)	DNC	1 (1)	2 (4)
Other	14 (8)	5 (6)	DNC	3 (3)	3 (6)
Susceptibility test method	Agar dilution	Agar dilution	Broth microdilution	Broth microdilution	Broth microdilution
Clinical breakpoints applied	CLSI M100 annual guidelines appropriate for the calendar yr	CLSI M100 2015 document	CLSI M100 2016 document	CLSI M100 2018 document	CLSI M100 2016 document
Antimicrobial susceptibility (% of susceptible isolates of E. anophelis, E. meningoseptica, and E. miricola, respectively)
Amikacin	0, 0, 8	DNC	0, 0, 9	6, 0, 73	DNC
Aztreonam	0, 0, 0	DNC	0, 0, 0	0, 0, 0	DNC
Cefepime	0, 0, 8	DNC	4, 0, 9	3, 0, 0	DNC
Ceftazidime	0, 0, 8	0, 0, 0	0, 0, 0	0, 0, 0	DNC
Ciprofloxacin	36, 23, 50	22, 23, 56	1, 0, 14	6, 43, 54	DNC
Colistin	0, 0, 0	DNC	0, 0, 0	0, 0, 0	DNC
Doxycycline	0, 67, 0	DNC	83, 91, 82	97, 100, 100	DNC
Gentamicin	0, 2, 8	22, 6, 45	0, 0, 0	1, 0, 82	DNC
Levofloxacin	71, 79, 92	29, 35, 100	16, 55, 77	48, 71, 82	DNC
Meropenem	0, 2, 8	DNC	0, 0, 0	0, 0, 0	DNC
Minocycline	100, 100, DNC	DNC	98, 100, 100	100, 100, 100	DNC
Piperacillin-tazobactam	93, 89, 92	92, 100, 94	73, 73, 73	3, 14, 27	DNC
Rifampin	DNC	96, 94, 66	DNC	81, 100, 27	DNC
Tigecycline	DNC	DNC	20, 55, 50	52, 100, 100	DNC
TMP-SMX	86, 89, 83	22, 6, 28	4, 0, 18	97, 100, 100	DNC
Tobramycin	0, 0, 0	DNC	0, 0, 0	0, 0, 0	DNC
Vancomycin	DNC	0, 0, 0	DNC	0, 0, 0	DNC

^aDNC, data not collected; TMP-SMX, trimethoprim-sulfamethoxazole; CLSI, Clinical and Laboratory Standards Institute; MALDI-TOF MS, matrix-assisted laser desorption ionization–time of flight mass spectrometry; RUO, research use only; GN, Gram negative.

To the best of our knowledge, this study is the largest report of antimicrobial susceptibility testing results of Elizabethkingia isolates from the United States.

Limitations of our study include the updates over time in identification methods and subsequent changes in the level of identification (i.e., species level or genus level). The identification of Elizabethkingia to the species level using MALDI-TOF MS is challenging; specifically, E. anophelis may be misidentified as E. meningoseptica depending on the database used (1, 2, 21, 23). Updates to MALDI-TOF MS libraries may allow more definitive species identification in the future. However, given the past and current challenges with the accurate identification of these organisms to the species level, it is not yet known whether there are distinct organism antimicrobial susceptibility patterns between species; further research into potential antimicrobial susceptibility differences between species may be useful. A final limitation is the unavailability of isolates for further assessment given the retrospective nature of this study.

Conclusion.

Elizabethkingia species exhibit variable MICs to antimicrobials; therefore, treatment of infections due to this genus should be guided by antimicrobial susceptibility testing results. Elizabethkingia MICs were within the susceptible range for 100% of isolates tested against minocycline and 90% tested against TMP-SMX. Based on the study findings, TMP-SMX and minocycline can be considered for guided therapy before antimicrobial susceptibility testing results are available. These results require correlation with in vivo data, and ongoing antimicrobial resistance surveillance and susceptibility testing of Elizabethkingia are necessary to support optimal treatment recommendations for the timely and appropriate management of infections.