Microbiologic and Clinical Implications of Bacteremia Due to Extended-Spectrum-β-Lactamase-Producing Klebsiella pneumoniae with or without Plasmid-Mediated AmpC β-Lactamase DHA-1

Abstract

Bacteremias caused by Klebsiella pneumoniae producing extended-spectrum beta-lactamase (ESBL-KP; n = 52) and producing both ESBL and AmpC-type DHA-1 beta-lactamase (ESBL-PMABL-KP; n = 20) were analyzed. Higher MIC₅₀s and MIC₉₀s for carbapenems, ciprofloxacin, and piperacillin-tazobactam were observed with ESBL-PMABL-KP than with ESBL-KP. Patients with oxyimino-β-lactam exposure and high modified Pitt bacteremia scores (HMPBSs) were at higher risk, while those with piperacillin-tazobactam and aminoglycoside exposure were at lower risk for ESBL-KP bacteremia. Patients with fluoroquinolone exposure, diabetes mellitus, and HMPBS were at higher risk, while those with aminoglycoside exposure were at lower risk, for ESBL-PMABL-KP bacteremia.

Enterobacteriaceae producing an extended-spectrum β- lactamase (ESBL) or a plasmid-mediated AmpC β-lactamase (PMABL) are resistant to broad-spectrum penicillins and oxyimino-β-lactams but usually remain susceptible to carbapenems (5, 11, 16). Strains producing PMABLs are also resistant to cephamycins. Flomoxef, a cephamycin unique in having a difluoromethylthioacetamido group at position 7, has better in vitro activity against ESBL-producing Enterobacteriaceae than other cephamycins (8) but is hydrolyzed by PMABLs (5, 8, 11).

In 2004, a cluster of in nosocomial bloodstream infections (BSIs) caused by flomoxef-resistant ESBL-producing Klebsiella pneumoniae isolates was noted at our hospital. In this work, we investigated those isolates and found that they also produced a PMABL. We also compared the clinical and microbiological features of bacteremic infections caused by ESBL plus PMABL-producing K. pneumoniae (ESBL-PAMBL-KP) with infections caused by K. pneumoniae producing only an ESBL (ESBL-KP) or lacking any of those enzymes.

ESBL production in K. pneumoniae was screened and then confirmed as recommended by the CLSI (1). Susceptibility testing was performed using the Etest (AB Biodisk, Solna, Sweden). Escherichia coli ATCC 25922 and K. pneumoniae ATCC 700603 were used as control strains. The breakpoints for susceptibility were ≤8 mg/liter for flomoxef (4) and according to the CLSI (2) for the other agents. As described previously, we identified the ESBL types using PCR amplification and sequencing for bla_SHV, bla_TEM, and bla_CTX-M (7), detected PMABL genes by use of multiplex PCR (13), and genotyped the K. pneumoniae isolates by use of pulsed-field gel electrophoresis (PFGE) (3). The PFGE profiles were interpreted as previously proposed (18).

The studied bacteremic patients belonged in two groups, group 1 (ESBL-KP bacteremia; n = 52) and group 2 (ESBL-PMABL-KP bacteremia; n = 20). A control group of patients with nosocomially acquired cefazolin-susceptible K. pneumoniae bacteremia (n = 100) from the same ward and the same period was included for comparison. Demographic and clinical information for patients of the three groups are reported in Table 1. Comparison of contingency data was carried out using the Chi-square test or Fisher's exact test, while comparison of continuous data was performed using Student's t test or the Mann-Whitney U test. For comparison to the control group, variables in group 1 and group 2 with P values of <0.1 in univariate analyses were separately entered into a multiple logistic regression model to identify the independent risk(s) for BSIs caused by ESBL-KP and ESBL-PMABL-KP, respectively. All comparisons were performed using SPSS 15 software for Windows (SPSS, Inc., Chicago, IL).

TABLE 1.

Demographic and clinical information of bloodstream infections due to K. pneumoniae^a

Variable	Value for patients			P
	Group 1 (n = 52)	Group 2 (n = 20)	Controls (n = 100)	Group 1 vs. controls	Group 2 vs. controls	Group 1 vs. group 2
Demographics
Male sex	30 (57.6)	13 (65.0)	60 (60.0)	0.729	0.804	0.657
Median age (range)	45 (22-91)	42 (30-91)	37 (18-91)	0.009	0.120	0.863
Clinical information
Neutropeniab	3 (5.7)	1 (5.0)	25 (25.0)	0.009	0.013	1
Malignancy	11 (21.2)	6 (30.0)	54 (54.0)	<0.001	0.085	0.537
Renal failurec	28 (53.8)	12 (60.0)	32 (32.0)	0.003	0.023	0.837
Diabetes mellitus	17 (32.6)	10 (50.0)	23 (23.0)	0.246	0.026	0.277
Hepatic dysfunctiond	16 (30.8)	4 (20.0)	30 (30.0)	1	0.428	0.535
CVC insertione	25 (48.1)	11 (55.0)	57 (57.0)	0.309	1	0.792
Foley catheter indwellinge	36 (69.2)	17 (85.0)	65 (65.0)	0.718	0.114	0.289
Mechanical ventilatory supporte	27 (51.9)	13 (65.0)	34 (34.0)	0.037	0.014	0.462
Median (range) modified Pitt bacteremia scoref	5 (0-11)	6 (1-10)	3 (0-10)	0.003	0.001	0.761
High modified Pitt bacteremia scoref (≥7 points)	16 (30.7)	11 (55.0)	10 (10.0)	0.003	<0.001	0.103
LOS ≧ 14 days before infection	40 (76.9)	15 (75.0)	63 (63.0)	0.027	0.442	1
Previous hospitalizationg	31 (59.6)	16 (80.0)	58 (58.0)	0.862	0.082	0.177
Recent antibiotic therapyh	44 (84.6)	17 (85.0)	76 (76.0)	0.302	0.564	1
Recent exposure to:
Carbapenemsi	11 (21.2)	6 (30.0)	19 (19.0)	0.831	0.364	0.537
Fluoroquinolonesj	14 (26.9)	10 (50.0)	18 (18.0)	0.214	0.007	0.113
β-Lactam/β-lactamase-inhibitork	18 (34.6)	6 (30.0)	47 (47.0)	0.168	0.219	0.786
Piperacillin-tazobactam	5 (9.6)	5 (25.0)	42 (42.0)	<0.001	0.625	0.127
Aminoglycosidesl	20 (38.5)	8 (40.0)	65 (65.0)	0.002	0.046	0.881
Oxyimino-β-lactamsm	37 (71.2)	14 (70.0)	45 (45.0)	0.003	0.051	0.847
Flomoxef	7 (13.4)	4 (20.0)	6 (6.0)	0.135	0.061	0.484
Appropriate empirical therapyn	30 (57.7)	5 (25.0)	100 (100)	<0.001	<0.001	0.026
Initial therapy failure at 72 ho	17 (33.3)	12 (60.0)	23 (23)	0.245	0.002	0.065
Mortality within 14 days	20 (38.5)	14 (70.0)	32 (32)	0.473	0.002	0.033

^aResults are shown for ESBL-producing K. pneumoniae without DHA-1 (group 1), ESBL-producing K. pneumoniae with DHA-1 (group 2), and non-ESBL-producing K. pneumoniae (controls). A patient was included once if culture of blood sampled at different time points all grew K. pneumoniae, and the first bloodstream infection (BSI) episode and the pathogen in this episode were counted. Values are numbers (percentages) of patients unless stated otherwise. Abbreviations: CVC, central venous catheter; ICU, intensive care unit; LOS, length of stay.

^bNeutropenia was defined as an absolute peripheral neutrophil count of <500 cells/μl.

^cRenal failure was defined as a serum creatinine level of >2.0 mg/dl or as kidney functions in patients receiving regular hemodialysis.

^dHepatic dysfunction was defined as high serum bilirubin level (>2.5 mg/dl) and alanine aminotransferase level (>80 IU/liter).

^eInvasive procedure was defined as central venous catheter implacement, endotracheal intubation, or urinary catheterization within 72 h before the emergence of K. pneumoniae BSI.

^fThe modified Pitt bacteremia score was calculated according to the following criteria: (i) oral temperature, 2 points for a temperature of ≤35°C or ≥40°C 1 point for a temperature of 35.1 to 36.0°C or 39.0 to 39.9°C, and 0 points for a temperature of 36.1 to 38.9°C; (ii) for hypotension, 2 points for an acute hypotensive event with decreases in systolic (>30 mm Hg) and diastolic (>20 mm Hg) blood pressures, use of intravenous vasopressor agents, or a systolic blood pressure of <90 mm Hg; (iii) for mechanical ventilation, 2 points; (iv) for cardiac arrest, 4 points; and (v) for mental status, 0 points for alert, 1 point for disoriented, 2 points for stuporous, and 4 points for comatose. A modified Pitt bacteremia score of ≥7 points was considered high (12).

^gPrior hospitalization was defined as a hospital stay within the 3 months before BSI developed.

^hAntibiotic exposure within 1 month.

ⁱIncluding imipenem, meropenem, and ertapenem.

^jIncluding ciprofloxacin, levofloxacin, and moxifloxacin.

^kIncluding amoxicillin-clavulanic acid, ampicillin-sulbactam, and piperacillin-tazobactam.

^lIncluding amikacin and gentamicin.

^mIncluding aztreonam, ceftazidime, cefotaxime, and ceftriaxone.

ⁿAppropriate empirical antibiotic therapy refers to an empirical antimicrobial treatment with the regimen containing at least one antibiotic to which the K. pneumoniae that subsequently grew from blood culture was susceptible in vitro.

^oThe initial response was considered favorable if clinical conditions and laboratory data improved; otherwise, it was regarded as initial therapy failure at 72 h.

PFGE patterns revealed multiclonality, indicating that each of the ESBL-KP and ESBL-PMABL-KP isolates was of an individual clone. One K. pneumoniae isolate might have more than one ESBL-encoding gene; a total of 62 ESBL-encoding genes were found in isolates of group 1, while 23 were found in those of group 2. CTX-M14 was predominant in group 2 (60.9% versus 30.6%; P = 0.022), while other ESBL enzymes, such as SHV-2, SHV-5, SHV-12, SHV-28, and CTX-M3 were not significantly different in each group. All ESBL-PMABL KP isolates produced the DHA-1 enzyme.

Compared to the ESBL-PMABL-KP isolates, the ESBL-KP isolates not harboring DHA-1 had significantly higher flomoxef susceptibility (98.1% versus 0%) and ciprofloxacin susceptibility (67.3% versus 10%) rates. While the ESBL-KP and ESBL-PMABL-KP isolates were universally susceptible to meropenem, a number of them were ertapenem intermediate (EI), ertapenem resistant (ER), imipenem intermediate (II), and/or imipenem resistant (IR). Specifically, among the 52 ESBL-KP isolates, 1 (1.9%) was EI and another 1 (1.9%) ER and II; among the 20 ESBL-PMABL-KP isolates, 14 (70%) were exclusively EI, 3 (15%) were exclusively ER, 2 (10%) were ER and II, and 1 (5%) was ER and IR. The MIC₅₀ and MIC₉₀ values of meropenem, imipenem, ertapenem, ciprofloxacin, and piperacillin-tazobactam for isolates in group 1 were overall lower than those for isolates in group 2 (Table 2).

TABLE 2.

Antimicrobial susceptibility of ESBL-producing K. pneumoniae isolates with or without DHA-1

Antibiotic	MIC (μg/ml) for ESBL-producing K. pneumoniae isolates
	Without DHA-1 (n = 52)			With DHA-1 (n = 20)
	Range	50%	90%	Range	50%	90%
Meropenem	0.015-0.5	0.03	0.06	0.06-1	0.12	0.5
Imipenem	0.12-2	0.25	0.5	0.25-4	0.5	2
Ertapenem	0.015-4	0.015	0.5	0.5-4	0.5	2
Ciprofloxacin	0.03->32	0.125	8	0.5->32	>32	>32
Amikacin	1->256	4	32	1->256	>256	>256
Piperacillin-tazobactam	0.25->256	8	32	0.25->256	8	64

Thirty patients (57.7%) in group 1 and five (25.0%) in group 2 received appropriate empirical antibiotic therapy (P = 0.026). Compared to what was found for group 1, there was a trend suggesting a higher percentage of initial therapy failure (60% versus 33.3%; P = 0.065) and a significantly higher 14-day mortality rate in group 2 (70% versus 38.5%; P = 0.033), which might result from the lower frequency of carbapenem-containing antibiotics (15% versus 32.7%) used on an empirical basis.

Upon separate comparison to the control group (Table 1), we found (i) significantly higher proportions of renal failure, mechanical ventilation, and high modified Pitt bacteremia scores (HMPBSs) and lower proportions of neutropenia and recent aminoglycoside exposure in group 1 and group 2, (ii) significantly older ages, longer hospital stays prior to bacteremia, higher proportions of recent exposure to an oxyimino-β-lactam(s), and lower proportions of underlying malignancy and recent piperacillin-tazobactam exposure in group 1, and (iii) significantly higher proportions of recent exposure to a fluoroquinolone(s) and underlying diabetes mellitus in group 2.

The multiple logistic regression model revealed that patients with recent oxyimino-β-lactam exposure (odds ratio [OR] = 3.22; 95% confidence interval [CI] = 1.45 to 7.45; P = 0.004) and HMPBS (OR = 3.13; 95% CI = 1.18 to 8.33; P = 0.022) were at higher risk, while those with recent piperacillin-tazobactam (OR = 0.17; 95% CI = 0.06 to 0.49; P = 0.001) and aminoglycoside exposure (OR = 0.31; 95% CI = 0.14 to 0.67; P = 0.003) were at lower risk for ESBL-KP bacteremia. On the other hand, patients with recent fluoroquinolone exposure (OR = 11.10; 95% CI = 2.64 to 46.61; P = 0.001), diabetes mellitus (OR = 4.64; 95% CI = 1.27 to 17.03; P = 0.021), and HMPBS (OR = 16.38; 95% CI = 4.00 to 67.05; P < 0.001) were at higher risk, while those with recent aminoglycoside exposure (OR = 0.22; 95% CI = 0.05 to 0.77; P = 0.019) were at lower risk for ESBL-PMABL-KP bacteremia.

This study highlighted some risk factors for bacteremia caused by ESBL-KP and ESBL-PMABL-KP isolates and the difference in antibiotic susceptibilities between these isolates. Previous reports suggested that substitution of piperacillin-tazobactam for expanded-spectrum cephalosporins reduced the prevalence of ESBL-producing Enterobacteriaceae in settings where ESBL is endemic (6, 10, 15) and that piperacillin-tazobactam could be clinically effective in the treatment of infections caused by ESBL-producing isolates showing MICs of the antibiotic of ≤16 μg/ml (14). Given this, it is not surprising to find that patients with prior exposure to piperacillin-tazobactam were reported to be at lower risk for acquisition of bacteremia caused by ESBL-producing Enterobacteriaceae (11, 19). Fluoroquinolone exposure has been a well-known risk factor for acquisition of infections due to ESBL-producing Enterobacteriaceae (11, 17, 20). Our data further indicated that patients with fluoroquinolone exposure were also at higher risk for ESBL-PMABL-KP bacteremia.

Additional remarkable susceptibility data in our series were the 100% resistance to flomoxef and the higher overall MIC values of ertapenem with the ESBL-PMABL-KP isolates than with K. pneumoniae isolates with ESBL alone. These findings might in part result from the acquisition of the bla_DHA-1 gene and from the presence of porin (e.g., OmpK36) deficiency in some isolates (9).

The findings that piperacillin-tazobactam and aminoglycoside-containing antibiotic regimens were associated with a lower risk for acquisition of ESBL-KP bacteremia and ESBL-PMABL-KP bacteremia, respectively, suggest that to minimize the emergence of infections caused by these multidrug-resistant microbes, piperacillin-tazobactam and/or aminoglycoside-containing regimens should be given priority over oxyimino-β-lactams and/or fluoroquinolones.