Clinical Correlation of the CLSI Susceptibility Breakpoint for Piperacillin- Tazobactam against Extended-Spectrum-β-Lactamase-Producing Escherichia coli and Klebsiella Species

Abstract

We assessed infections caused by extended-spectrum-β-lactamase-producing Escherichia coli or Klebsiella spp. treated with piperacillin-tazobactam to determine if the susceptibility breakpoint predicts outcome. Treatment was successful in 10 of 11 nonurinary infections from susceptible strains and in 2 of 6 infections with MICs of >16/4 μg/ml. All six urinary infections responded to treatment regardless of susceptibility.

Treatment of infections caused by extended-spectrum-β-lactamase (ESBL)-producing Escherichia coli, Klebsiella pneumoniae, or Klebsiella oxytoca with a cephalosporin is associated with treatment failure and increased mortality (11-13, 19, 20, 25). The Clinical Laboratory Standards Institute (CLSI) (formerly NCCLS) recommends reporting ESBL-producing strains of E. coli and Klebsiella spp. as resistant to all penicillin, true cephalosporin, and monobactam antimicrobial agents but recommends reporting ESBL-producing isolates as susceptible to β-lactam/β-lactamase inhibitor combination antimicrobials, like piperacillin-tazobactam, when they test as such in the laboratory (4). However, reports of clinical failures have led some to recommend a carbapenem antimicrobial for all ESBL infections (19, 25). Thus, regardless of demonstrable in vitro susceptibility, many clinicians are reluctant to treat infections caused by ESBL producers with piperacillin-tazobactam. The purpose of this study was to determine if the current CLSI susceptibility breakpoint for piperacillin-tazobactam (≤16/4 μg/ml) reliably predicts treatment outcome. (This study was presented in part at the 42nd Annual Meeting of the Infectious Diseases Society of America, Boston, Mass., October 2004 [abstr. 440].)

Study patients were identified by retrospective review of databases at participating laboratories, recruited from ClinMicroNet (an electronic information network of leading clinical microbiology laboratory directors), from January 2001 to December 2003. Patient data were collected on age, gender, body weight, laboratory studies (sites of infection; antimicrobial regimen), and comorbid illnesses. Skin and soft-tissue specimens were included when obtained at surgery. Outcomes were determined for patients with infections caused by ESBL producers treated with piperacillin-tazobactam, alone or in combination with other antimicrobials, for a minimum of 72 h. Outcomes determined on clinical and/or bacteriologic parameters were compared to results of susceptibility testing by the participating laboratories. Clinical cure was defined as improvement in clinical status and the physician's opinion as to treatment success or failure. Bacteriologic cure was defined as negative cultures after ≥72 h of treatment. Treatment failure was defined as lack of clinical improvement, death from infection, or persistently positive cultures. Statistical analysis was by Fisher's exact test. Approval for this study was obtained from the respective institutional review boards.

Bacterial identification was performed by conventional methods. Susceptibility testing was performed by microtube broth dilution or disk diffusion, and the actual or extrapolated MICs recorded (4, 10, 16, 17). Where quantitative MICs were not recorded, isolates reported as “susceptible” or “resistant” were assigned MICs of ≥16/4 μg/ml or >128/4 μg/ml (4).

The percentage of time above the MIC (%T>MIC) was calculated using a formula that combined individual pharmacokinetic parameters (piperacillin-tazobactam regimen, dosing interval, and MIC) plus published values for fraction unbound (70%) drug, volume of distribution (0.15 liter/kg), and t_1/2 (0.75 h) of piperacillin-tazobactam: %T>MIC = ({ln[(fu × dose)/(V × MIC)] × [t_1/2/ln(2)]} × 100)/τ (where ln = natural logarithm, fu = fraction unbound, V = volume of distribution [liters/kg], t_1/2 = elimination half-life [h], and τ = dosing interval [h]) (5, 9, 18, 26, 29). A one-compartment, first-order, intravenous model was used to calculate %T>MIC. While there is disagreement as to the accuracy of a one-compartment model for pharmacodynamic analysis (30), the infusion time of piperacillin-tazobactam is relatively uniform, which negates much of the differences (13, 29). Tazobactam exposure was estimated by calculating the area under the concentration-time curve (AUC_0-24).

We identified 148 patients with infections caused by ESBL-producing E. coli (62 patients; 42%), K. pneumoniae (72 patients; 49%), or K. oxytoca (13 patients; 9%) or mixed E. coli and K. pneumoniae infection (1 patient). The majority of ESBL-producing isolates were from urine (52%), followed by blood (18%), sputum (17%), skin and soft-tissue (11%), abdominal/peritoneal (3%), and additional sites (3%). Most patients were treated with a single antimicrobial agent (83%). Fluoroquinolones were the most common antimicrobials (30%), followed by carbapenems (18%), piperacillin-tazobactam (16%), other β-lactams (9%), ampicillin-sulbactam (7%), aminoglycosides (6%), and trimethoprim-sulfamethoxazole (6%).

Twenty-three patients were treated with piperacillin-tazobactam alone or in combination with other antimicrobials for at least 72 h (E. coli, 9 patients; K. pneumoniae, 13 patients; and K. oxytoca, 1 patient), 17 of which had non-urinary-tract infections (UTI) and 6 of which had UTI. There were 18 males and 5 females (median age, 65 years; range, 47 to 87 years).

MICs of piperacillin-tazobactam for the 23 ESBL producers ranged from <0.5/4 to >128/4 μg/ml. Outcomes of non-UTI and UTI were analyzed separately. Piperacillin-tazobactam treatment was successful in 91% (10 of 11) of non-UTI caused by ESBL producers with MICs of ≤16/4 μg/ml but in only 20% (1 of 5) caused by ESBL producers with MICs of >16/4 μg/ml (Fig. 1) (P = 0.027). Piperacillin-tazobactam treatment was successful in 100% (6 of 6) of UTI irrespective of MIC (P = 1.0).

FIG. 1.

Outcome of treatment (for those patients receiving piperacillin-tazobactam alone or with another agent for less than 48 h) based on the in vitro MIC.

Pharmacokinetic/pharmacodynamic data were available for 15 patients. Treatment was successful in 90% (9 of 10) of patients with non-UTI when %T>MIC was greater than 40% (Fig. 2). Calculated tazobactam AUC_0-24s for patients were in excess of 96 mg/h/liter, the target concentration of tazobactam proposed by Dudley and by Ambrose and colleagues where bacteria are rendered functionally β-lactamase negative for the duration of the dosing interval (2, 7). In contrast, treatment was unsuccessful for both patients with non-UTI where %T>MIC was less than 40% (Fig. 2). Despite favorable pharmacokinetic/pharmacodynamic estimates, treatment was unsuccessful for one patient with bacteremia and cholangitis caused by a susceptible ESBL-producing E. coli isolate (MIC, 16/4 μg/ml). Treatment was unsuccessful for four patients with infections from non-susceptible ESBL-producing Klebsiella spp. Three successfully treated patients received antimicrobials to which the isolates were susceptible in addition to piperacillin-tazobactam (gentamicin, two patients; ampicillin-sulbactam, one patient). However, all three received ≤48 h of the second agent, with likely little impact on outcome. Thus, 87% of patients received piperacillin-tazobactam with less than 2 days of a second antimicrobial.

FIG. 2.

Outcome of treatment (for those patients receiving piperacillin-tazobactam alone or with another agent for less than 48 h) based on the percentage time above the MIC.

Although clinical outcome was not determined for other antimicrobials, data were available on four patients treated with meropenem. Meropenem treatment was successful for two patients with infections caused by piperacillin-tazobactam-resistant K. pneumoniae pneumonia and bacteremia but failed in another two patients with K. oxytoca and K. pneumoniae pneumonias.

Although carbapenem treatment has previously been associated with a more favorable clinical outcome, it has been accompanied by the emergence of carbapenem resistance and increases in Acinetobacter and Stenotrophomonas infections (1, 8, 15, 21, 22). In contrast, treatment with piperacillin-tazobactam has been associated with successful control of outbreaks caused by ESBL producers without loss of susceptibility (3, 12, 20, 21, 23, 24). Similarly to earlier reports, in the present study, occasional piperacillin-tazobactam and carbapenem failures were demonstrated in apparently susceptible isolates.

The greater in vitro sensitivity of piperacillin-tazobactam to the inoculum effect of ESBL producers, particularly SHV-derived ESBLs, has been blamed for piperacillin-tazobactam failure in infections caused by susceptible isolates (28). However, in earlier small case series, treatment failures were seen with susceptible isolates from blood (n = 2), urine (n = 2), and respiratory secretions (n = 2), sites not classically associated with high inocula (3, 20, 21, 28). Studies now suggest that the inoculum effect is an artifact of in vitro susceptibility testing and of little clinical relevance (6). Notably, pharmacokinetic/pharmacodynamic analyses were not performed in earlier studies, whereas the present study predicted successful outcome in greater than 90% of our patients with non-urinary isolates when %T>MIC exceeded 40%.

The literature on piperacillin-tazobactam is divided regarding the pharmacodynamic role of the tazobactam component (2, 14, 27). In the present study, the calculated tazobactam AUC_0-24 was in excess of the estimated concentration required for β-lactamase activity throughout the dosing interval (2, 7). Thus, we agree with Johnson and colleagues that dose adjustments can be based on pharmacokinetics of piperacillin alone (9).

In conclusion, UTI caused by piperacillin-tazobactam-susceptible ESBL producers may be successfully treated, as expected. Piperacillin-tazobactam treatment was also successful for 90% of non-UTI caused by ESBL producers with MICs less than the CLSI susceptibility breakpoint.