Nosocomial Outbreak of Klebsiella Infection Resistant to Late-Generation Cephalosporins

Methods

Nosocomial Isolates

During routine infection-control surveillance of multi-resistant nosocomial pathogens, a cluster of CRKP isolates was noted in November 1989. Worksheets in the Clinical Microbiology Laboratory were reviewed retrospectively, and the first such isolate was noted to have occurred in October 1988. Subsequently, information on all CRKP isolates was collected retrospectively from November 1989 to October 1988 and prospectively from December 1989 to October 1990.

Biochemical identification required a gram-negative, nonmotile bacillus that fermented lactose with gas production, did not produce H_2S, was indole negative, and metabolized citrate and/or malonate. Confirmation with AP1-20E (API Analytab Products; Plainview, New York) or Vitek (Vitek Systems, McDonnel-Douglas; Hazelwood, Missouri) was obtained whenever the above biochemical profile was associated with a susceptibility pattern other than the typical ampicillin and carbenicillin resistance of Klebsiella species.

Clinical Data

Clinical data were obtained from the charts of all patients harboring CRKP from October 1988 through February 1990. Clinical assessment was determined according to the 1988 Centers for Disease Control and Prevention definitions for nosocomial infections [34].

Susceptibility Testing

Susceptibility of clinical isolates was tested initially by the Clinical Microbiology Laboratory according to the National Committee for Clinical Laboratory Standards (NCCLS) using the Kirby-Bauer disc diffusion method on Mueller-Hinton agar (BBL Microbiology Systems; Cockeysville, Maryland). Ceftazidime resistance was defined as a zone of inhibition less than 15 mm; susceptibility as greater than 17 mm; and intermediate susceptibility as 15 to 17 mm. Strains demonstrating intermediate susceptibility were included with resistant strains for clinical and microbiological analysis. Ceftizoxime was used as a class representative of all other third-generation cephalosporins. Disc diffusion and macrodilution susceptibility tests were performed by the Infectious Disease Research Laboratory on random CRKP isolates obtained after the outbreak was recognized. In these studies, individual third-generation cephalosporins were tested. Broth macrodilution minimal inhibitory concentrations (MIC) and minimal bactericidal concentrations (MBC) were performed according to NCCLS guidelines [35].

Antimicrobial Agents

Antimicrobial agents were obtained as follows: ticarcillin and potassium clavulanate, Beecham Laboratory, Bristol, Tennessee; ceftazidime, Fujisawa SmithKline Corporation, Philadelphia, Pennsylvania; cefuroxime and ceftazidime, Glaxo Pharmaceuticals, Research Triangle Park, North Carolina; cefoxitin, Hoechst Roussel Pharmaceuticals, Inc., Somerville, New Jersey; cefoxitin and imipenem, Merck Sharp & Dohme, West Point, Pennsylvania; cefotetan, Stuart Pharmaceuticals, Wilmington, Delaware; cefmetazole, The Upjohn Company, Kalamazoo, Michigan; Cefepime, Bristol Meyers Company, Syracuse, New York: piperacillin and tazobactam, Lederle Pharmaceuticals, Pearl River, New York: and moxalactam, Eli Lilly & Company, Indianapolis, Indiana.

Results

Epidemiology

From October 1988 to April 1990, 432 isolates of CRKP were recovered from 155 patients, representing 17.3% of all Klebsiella pneumoniae isolates. The peak incidence rate of 72 patients per 1000 average daily census was reached in November 1989 (Figure 1). This represented a prevalence of 25 colonized or infected patients, or 30% of all those from whom Klebsiella pneumoniae had been isolated. Barrier precautions were instituted on colonized and infected patients, and increased restriction of ceftazidime was implemented. Subsequently, the incidence rate of CRKP among medical and surgical patients declined by approximately 60% within the following year.

Patients were colonized or infected with CRKP after a mean of 37.2 days of hospitalization. Ten percent of patients were identified during the first week, 20% during the second week, 22% during the third and fourth weeks, 30% from the fifth to eighth weeks, and 18% after the eighth week of hospitalization. Thus, 70% of patients were identified after the second week of hospitalization.

Clinical records of 133 of the 142 patients harboring CRKP through February 1990 were available for analysis. Of these, 52 (39%) met criteria for CRKP infection in 68 separate sites (Table 1). Infections in surgical patients occurred primarily in the surgical intensive care unit. Among medical patients, infections occurred both on general wards and in the intensive care unit. Eighty-four percent of bacteremias and lower respiratory tract infections occurred in both intensive care units, whereas 66% of urinary infections occurred on the wards. Conversely, 73% of CRKP ward infections were urinary, and 62% of intensive care unit CRKP infections were bacteremic or pulmonary.

The average age of CRKP-infected patients was 70 years. All suffered from various acute and chronic disorders, most of which were rapidly or ultimately fatal. No patient had an absolute neutrophil count less than 1000/mm³.

Previous Antibiotic Use and Ceftazidime-resistant Klebsiella pneumoniae Isolation

Sufficient records of antibiotic administration were available for 127 of 142 patients identified through February 1990. Before CRKP was isolated, 91 of 127 patients (72%) received more than 7 days of antibiotic therapy. An average of 4.7 antibiotics was administered per patient before CRKP was isolated. Ceftazidime had been used previously in 41% of all patients from whom CRKP was isolated. However, its use preceded CRKP isolation in 76% of those with pneumonia and in 100% of those with bacteremic pneumonia (five patients).

Microbiology and Antibiotic Susceptibility

Sixty-two CRKP isolates showed intermediate susceptibility by disc diffusion, and 374 isolates showed complete resistance to ceftazidime. Table 2 lists the susceptibilities of all 436 CRKP isolates identified by the Clinical Microbiology Laboratory from October 1988 to April 1990. Nearly 100% of isolates were resistant to mezlocillin, tobramycin, gentamicin, and tetracycline. More than 90% of isolates were susceptible to ciprofloxacin, cefotetan, and ceftizoxime; all isolates were susceptible to imipenem by disc diffusion testing. Susceptibility to cefazolin and cefoxitin was found in 31.9% and 76.8% of CRKP, respectively. Amikacin resistance was present in 53% of CRKP in contrast to 8% of ceftazidime-susceptible Klebsiella pneumoniae.

Disc diffusion susceptibility studies were repeated by the Infectious Disease Research Laboratory on 32 randomly selected CRKP isolates (Table 3). Fifteen -lactam agents were tested. Only imipenem and piperacillin-tazobactam yielded inhibitory zones uniformly above the NCCLS breakpoint. Inhibitory zones for all other antibiotics varied widely, with moxalactam, ceftizoxime, cefotetan, cefoxitin, and ceftriaxone inhibiting more than 90% of isolates. Macrodilution assays showed MIC₉₀ of 64 mg/L or more for all antibiotics tested except imipenem, cefotetan, and moxalactam. However, the MBC₉₀ for cefotetan and moxalactam was greater than 64 mg/L compared with 16 mg/L for imipenem (Table 4).

Mechanism of Ceftazidime Resistance

Studies that will be reported separately have shown several plasmids and at least five different -lactamases in a randomly selected group of nine CRKP isolates. The -lactamase responsible for ceftazidime hydrolysis had a pI of 5.6 and was common to all resistant strains [36]. Enzymatic characteristics were consistent with TEM-10 and the more recently described TEM-26. Both enzymes have been found in limited clusters associated with previous ceftazidime therapy [27, 33]. In each situation, susceptibility to cephamycins, cefotaxime, and ceftriaxone was reported. However, the bactericidal activity of these agents against ceftazidime-resistant isolates was not measured.

Therapy and Outcome

Therapy and outcome data were available for 43 patients. All 13 patients receiving no therapy directed against CRKP died. An additional 10 deaths occurred among treated patients, resulting in a crude mortality rate of 53%. Treatment regimens that included imipenem, in contrast to other antibiotics, yielded the most favorable results. Third-generation cephalosporins were effective in only two patients, both with nonbacteremic bacteriuria. Cefotetan was not used to treat any CRKP infection.

Infection-Control Intervention and Ceftazidime Restriction

Yearly ceftazidime consumption increased 600% during the 2 years immediately preceding recognition of the CRKP outbreak (Figure 2). Despite a program in which the use of late-generation cephalosporins required approval by the Infectious Disease Section, ceftazidime use increased in relation to an outbreak of infections caused by multiresistant Acinetobacter calcoaceticus, usually susceptible only to ceftazidime, amikacin, and imipenem. After the CRKP outbreak was recognized, ceftazidime approvals were reduced by 80%, leading to an increase in the use of imipenem. The first 6 months of ceftazidime restriction resulted in a 73% reduction in grams purchased. After the institution of barrier precaution for colonized and infected patients, and curtailment of ceftazidime use starting in November 1989, the incidence rate of CRKP progressively decreased. This occurred without clustering patients or personnel into cohorts and without attempts to eradicate colonization from either population.

Discussion

Descriptions of plasmid-mediated, extended-spectrum -lactamases have proliferated with the increasing recognition of CRKP isolates during the past 5 years. Major outbreaks have occurred in France, and three limited outbreaks have been described in the United States [31-33]. Our experience documents the capacity of CRKP to become established and cause disease throughout a general medical and surgical population under the selective pressure of ceftazidime use despite an active antibiotic control program. The results of increased ceftazidime restriction and infection-control measures have shown that such an outbreak can be controlled. However, following increased imipenem use for the treatment of CRKP infections Figure 2, imipenem-resistant Acinetobacter infections emerged in our institution [37].

Previous reports of CRKP have suggested that such strains may be susceptible to third-generation cephalosporins other than ceftazidime. Our routine disc diffusion susceptibility studies confirmed these findings. However, macrodilution susceptibility tests against standard inocula revealed an almost uniform pattern of resistance to all cephalosporins and cephamycins, with the exception of moxalactam and cefotetan, which were inhibitory but not bactericidal. Thus, disc diffusion susceptibility standards for CRKP did not correlate well with acceptable minimal inhibitory and bactericidal concentrations.

The multiple-drug-resistant nature of CRKP may impair the responsiveness of an outbreak to the restriction of a single antibiotic. For instance, the continued use of amikacin, to which many of our CRKP were resistant, may have exerted selective pressure to preserve amikacin-resistant CRKP. Indeed, despite modification of ceftazidime use in the Massachusetts outbreak, CRKP decreased transiently, and after 7 months, resistant isolates persisted [31]. Another troublesome feature of our outbreak was the high proportion of isolates resistant to the bactericidal action of cefotaxime, ceftizoxime, ceftriaxone, and the cephamycins: cefoxitin, cefmetazole, and moxalactam. In previous reports, CRKP isolates were considered susceptible to other late-generation cephalosporins and cephamycins, although bactericidal activity was not documented [27, 28, 33]. Jacoby and Medieros [26] have noted that the MICs of cefotaxime and ceftriaxone increase dramatically as the inoculum of CRKP is increased, a finding consistent with poor bactericidal effect. In contrast, they considered cephamycins and imipenem to be acceptable treatment options. However, our isolates were more susceptible to the bactericidal activity of imipenem than that of cephamycins.

Alternative mechanisms of possible cephamycin resistance in Klebsiella have come to light in recent publications. Unrelated classes of antibiotics, such as the quinolones, may select for deficiency of an ompF-like 39-kd outer-membrane protein that may be responsible for -lactam transport in Klebsiella [38]. In a separate report, deficiency of a 40-kd outer-membrane protein in a TEM-3-producing strain of Klebsiella pneumoniae seems to have mediated a 4- to 16-fold increase in the MIC to all -lactam antibiotics, with the exception of imipenem [39]. The final MICs showed susceptibility only to imipenem, cefotetan, and moxalactam, as seen in isolates from this outbreak. Such a combination of resistance mechanisms was recently described in Escherichia coli [40]. Finally, the acquisition of a plasmid-located class I -lactamase (named MIR-1) has been identified in Klebsiella pneumoniae isolates resistant to all clinically used -lactams except imipenem [32]. A sequence of the MIR-1 gene demonstrated 90% identity with the ampC gene of Enterobacter cloacae. Thus, the outbreak we describe may be due to multiple resistance mechanisms acting in concert.

The data described above have several clinical implications. Disc diffusion susceptibility standards must be re-evaluated in the context of extended-spectrum, plasmid-mediated cephalosporin resistance. As noted by others, class representatives of antibiotics cannot be substituted for individual agents [28, 33]. Plasmid-mediated TEM -lactamase resistance of Klebsiella to all late-generation cephalosporins may escape detection except by ceftazidime susceptibility testing, particularly when the disc diffusion method is used. In our experience, the outbreak escaped detection for approximately 1 year because most CRKP strains appeared sensitive to other late-generation cephalosporins by disc diffusion susceptibility. Despite relatively low MICs of certain late-generation cephalosporins and cephamycins against our CRKP isolates by tube dilution assay, only imipenem exhibited consistent bactericidal activity. Thus, we believe that imipenem is the preferred therapy for serious CRKP infections.

Finally, the combination of extended-spectrum cephalosporin and aminoglycoside resistance illustrates the remarkable genetic potential conferred by the presence of multiple plasmids, each with multiple resistance determinants, in a single pathogenic species. Control measures other than restriction of specific antibiotics must be explored for the containment and elimination of multiple-drug-resistant organisms. Our experience illustrates that such resistance is controllable by combined antibiotic restriction and infection-control measures. Subsequent to the described study period, the incidence of CRKP has remained at a constant low level.