Piperacillin-Tazobactam Versus Carbapenems for the Treatment of Nonbacteremic Urinary Tract Infections due to Extended-Spectrum Beta-Lactamase-Producing Enterobacteriaceae

Abstract

Background: Carbapenems are the drug of choice for treatment of infections due to extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae. Current evidence regarding piperacillin-tazobactam (PTZ) as an effective treatment alternative remains controversial. The purpose of this study was to determine the efficacy of PTZ versus carbapenems for treatment of nonbacteremic urinary tract infections (UTIs) due to ESBL-producing Enterobacteriaceae. Methods: A retrospective cohort study of patients treated for ESBL-related UTIs was conducted at three medical centers in the greater Middle Tennessee area. Patients were included if they were ≥ 18 years old, had a positive urine culture with an ESBL-producing organism, and received ≥ 48 hours of treatment with PTZ or carbapenem. Patients with bacteremia as well as those with isolates resistant to the treatment regimen selected were excluded. The primary objective was to determine the difference in clinical response between PTZ and carbapenem for treatment of ESBL-related UTIs. Clinical response was defined as absence of all of the following: (1) repeat admission for UTI caused by the same organism within 6 months, (2) repeat urine culture within 6 months showing growth of the same organism, or (3) a change in antimicrobial regimen due to subjective failure as determined by the ordering provider. Results: A total of 180 patients were included in the analysis (PTZ = 39; carbapenem = 141). There was no difference in clinical response between patients receiving PTZ and carbapenem (74.4% versus 80.9%; P = .38). Conclusion: PTZ may be an effective alternative to carbapenems for treatment of nonbacteremic UTIs due to ESBL-producing Enterobacteriaceae.

Introduction

Antimicrobial resistance is a key healthcare threat worldwide. The emergence of bacterial resistance to broad-spectrum beta-lactam antimicrobials is increasing.¹ One of the major mechanisms of resistance to beta-lactam agents is the production of extended-spectrum beta-lactamases (ESBLs), and incidence of this is also on the rise.¹ The spread of ESBL genes has been identified among the Enterobacteriaceae family, of which the most common clinically encountered are Escherichia coli, Proteus mirabilis, and Klebsiella species.¹ Currently, carbapenems are considered the drug of choice for treatment of infections due to ESBL-producing isolates. However, careful consideration must be given to utilization of these broad-spectrum antimicrobials, as high empiric use of carbapenems has allowed for development of carbapenem-resistant Enterobacteriaceae, for which treatment options are sparse.^1,2

Current treatment alternatives to carbapenems for ESBL infections are limited.^1,2 While newly approved agents such as ceftazidime/avibactam, ceftolozane/tazobactam, and meropenem/vaborbactam are available options, their use is limited by cost, procurement, timely susceptibility testing, and minimal clinical experience for a variety of indications.³ Due to their broad spectrum of activity, reservation of these agents for more resistant infections (e.g. carbapenem-resistant Enterobacteriaceae) is recommended.

Although some ESBL isolates may be categorized as susceptible to beta-lactam/beta-lactamase inhibitors (BL/BLI) such as piperacillin-tazobactam (PTZ) according to laboratory testing, evidence regarding their in vivo clinical efficacy is controversial.³ While some studies have shown worse outcomes with PTZ therapy when compared to carbapenems for treatment of documented ESBL infections,^4,5 other studies have indicated PTZ may be a viable alternative.^6-9 A systematic review of carbapenem-sparing therapy against ESBL bloodstream infections concluded that complete avoidance of carbapenems cannot be justified and that consideration of PTZ as an alternative treatment depends heavily on the source of infection.¹⁰ It has been proposed that utilization of PTZ for treatment of ESBL infections of mild to moderate severity should be considered, with infections of a urinary source being a primary area of discussion.^3,10,11 Because of the high proportion of ESBL infections that are of a urinary source, evidence and guidelines for optimal treatment of these infections is of the utmost importance to allow for the preservation of carbapenems.

The purpose of this study is to determine the efficacy of piperacillin-tazobactam compared with carbapenems for treatment of nonbacteremic urinary tract infections (UTIs) caused by ESBL-producing Enterobacteriaceae.

Methods

This was a multicenter, IRB-approved, retrospective cohort study conducted at three medical centers associated with one health system in the greater Middle Tennessee area. The charts reviewed were identified through a microbiology lab report that recognized all inpatient encounters during the time period of January 1, 2014, to June 30, 2016, that had positive urine cultures with ESBL-producing Enterobacteriaceae. The researchers were located at one of the medical centers and were able to access data from the other medical centers within the same health system via use of a shared electronic medical record (EMR). IRB approval was obtained from the health system which covers research conducted at all three medical centers included in the study.

To be included in the study, patients had to be 18 years of age or older, admitted to one of three medical centers, have a UTI due to ESBL-producing Enterobacteriaceae confirmed by microbiology urine culture reports, and have received at least 48 hours of treatment with either PTZ or a carbapenem. Of note, UTI was confirmed by physician documentation and decision to treat. If asymptomatic bacteriuria was documented by the physician, it was recorded. Formulary carbapenem agents at the medical centers included in the study were meropenem and ertapenem. Patients were excluded from the study if the ESBL-producing organism was resistant to the antimicrobial used for treatment or if bacteremia was present. If a patient had multiple admissions during the study period, only the first patient encounter that met study criteria was included in the analysis. Cases were categorized as one of two groups based on the definitive antimicrobial therapy received: PTZ or carbapenem. Definitive antimicrobial therapy was defined as the antimicrobial therapy utilized after organism and susceptibility information was reported. A pharmacist-led renal dosing protocol was in place for both PTZ and carbapenems at each of the medical centers included in this study; dosing strategies were uniform across medical centers and were reviewed daily. Of note, pharmacodynamic dosing of PTZ and meropenem was implemented during the study period at all three medical centers simultaneously and remained under pharmacist-led dosing.

The primary outcome of the study was to determine the difference in clinical response between PTZ and carbapenems for the treatment of nonbacteremic UTIs due to ESBL-producing Enterobacteriaceae. Cases were deemed to have a clinical response if all of the following criteria were absent: (1) repeat admission for UTI caused by the same organism within 6 months of initial diagnosis, (2) repeat urine cultures within 6 months of initial diagnosis showed positive growth for the same organism, or (3) a change in antimicrobial regimen occurred due to subjective clinical failure as determined by the ordering provider. Secondary outcomes included duration of treatment, length of hospital stay, and inpatient mortality.

Minimum inhibitory concentration (MIC) data were collected as determined by Vitek2 on PTZ cases specifically as the assumption was that MIC data for carbapenem cases would be inconsequential.

The primary outcome of clinical response, which is a nominal, dichotomous variable, was analyzed using a Chi-square test. Secondary outcomes and background characteristics were analyzed by appropriate statistical tests, with t test being used for continuous variables and Chi-square tests being used for nominal variables. Statistical analyses were completed by the study authors utilizing JMP^® Pro v11.2.0 (SAS Institute Inc., Cary, NC).

Results

A total of 679 patient encounters were screened (Figure 1). Of these, 180 cases met study criteria. The most common reasons for exclusion were definitive therapy of less than 48 hours, lack of treatment with PTZ or a carbapenem, or concomitant bacteremia present. In addition, some patients had multiple encounters within the study period and according to the study plan, only the first patient encounter meeting inclusion/exclusion criteria was accepted. Of the 180 patient encounters included in the final analysis, 39 received definitive therapy with PTZ and 141 received definitive therapy with a carbapenem (105 with meropenem and 36 with ertapenem).

Figure 1.

Patient enrollment.

^aPiperacillin-tazobactam.

Baseline characteristics were overall similar between the two treatment groups, with the majority of patients being female and mean age of approximately 71 years old (Table 1). Approximately 62% of patients in the PTZ group required admission to the intensive care unit (ICU) at some point during their hospital stay compared to 40% of patients in the carbapenem group. Table 2 describes the microbiologic findings between the two groups with Escherichia coli representing the majority of organisms isolated, followed by Klebsiella pneumoniae. Comorbidities in general and urinary-related comorbidities specifically were also similar between the two groups.

Table 1.

Baseline Characteristics.

	PTZ (n = 39)	Carbapenem (n = 141)	P value
Female, n (%)	25 (64.1%)	106 (75.2%)	.22
ICU admission, n (%)	24 (61.5%)	56 (39.7%)	.02
Age (years), mean ± SD	71.4 ± 16.9	71.1 ± 15.1	.90
Weight (kg), mean ± SD	79.8 ± 21.3	81.3 ± 25.7	.71
Tmaxa (°C), mean ± SD	37.4 ± 0.83	37.3 ± 0.94	.56
WBCb (×109/L), mean ± SD	12.0 ± 6.5	12.1 ± 7.0	.88
Comorbidities
Immunosuppressive agents	1 (2.6%)	9 (6.4%)	.69
Solid organ transplant	0 (0%)	5 (3.6%)	.59
Renal disease	12 (30.8%)	51 (36.2%)	.58
Diabetes	14 (35.9%)	67 (47.5%)	.21
Neurogenic bladder	5 (12.8%)	11 (7.8%)	.34
Benign prostatic hyperplasia	5 (12.8%)	10 (7.1%)	.32
History of urinary	2 (5.1%)	9 (6.4%)	1
incontinence	5 (12.8%)	16 (11.4%)	.78
Chronic catheter	12 (30.8%)	50 (35.5%)	.70
History of UTIs	3 (7.7%)	26 (18.4%)	.14

Note. PTZ = piperacillin-tazobactam; ICU = intensive care unit; WBC = White blood cell; UTIs = urinary tract infections.

^aMaximum temperature at presentation (date of urine culture collection)

^bWBC count at presentation (date of urine culture collection)

Table 2.

Microbiology and Urinary-Related Characteristics.

	PTZ (n = 39)	Carbapenem (n = 141)	P value
Organism isolateda, n (%)
Escherichia coli	36 (92.3%)	111 (78.7%)	.06
Klebsiella pneumoniae	4 (10.3%)	30 (21.3%)	.17
Klebsiella oxytoca	0 (0%)	1 (0.71%)	1
Proteus mirabilis	0 (0%)	1 (0.71%)	1
Urinary instrumentation, n (%)
Foley catheter	12 (30.8%)	47 (33.3%)	.85
Intermittent catheter	3 (7.7%)	5 (3.6%)	.37
Otherb	4 (10.3%)	8 (5.7%)	.29
No instrumentation	22 (56.4%)	84 (59.6%)	.72
Documented UTI characteristics, n (%)
Pyelonephritis	0 (0%)	10 (7.1%)	.12
Prostatitis	0 (0%)	1 (0.71%)	1

Note. PTZ = piperacillin-tazobactam; UTI = urinary tract infection.

^aSome patients had urine cultures with multiple organisms.

^bIncludes stents, urostomy, nephrostomy, and bladder sling.

No significant difference was observed between PTZ and carbapenem for the primary outcome as clinical response occurred in 74.4% and 80.9% of patients, respectively (P = .38) (Figure 2). Furthermore, the groups did not differ when comparing the incidence of repeat admissions for the same UTI within 6 months, repeat positive urine cultures with the same organism within 6 months, or change in antimicrobial regimen due to subjective failure as determined by the ordering provider (Table 3).

Figure 2.

Primary outcome.

Table 3.

Criteria Met for Lack of Clinical Response.

	PTZ (n = 39)	Carbapenem (n = 141)	P value
Repeat admission for the same UTIa	6 (15.4%)	28 (19.9%)	.65
Positive repeat urine culturesa	6 (15.4%)	21 (14.9%)	1
Change in antimicrobial regimen occurred	1 (2.6%)	3 (2.1%)	1

Note. PTZ = piperacillin-tazobactam; UTI = urinary tract infection.

^aWithin 6 months.

The majority of isolates treated with PTZ had a piperacillin MIC of ≤ 4 μg/mL (Table 4). All of the cases where isolates had piperacillin MICs of 16 μg/mL and 32 μg/mL reported met criteria for clinical response.

Table 4.

MIC Data.

Piperacillin MICa (μg/mL)	PTZa (n = 38)b
≤ 4	28 (73.7%)
8	7 (18.4%)
16	1 (2.6%)
32	2 (5.3%)

Note. MIC = minimum inhibitory concentration; PTZ = piperacillin-tazobactam.

^aMIC was determined using Vitek2.

^bMIC for one isolate was not reported.

There was no significant difference in the mean duration of antimicrobial therapy (5.0 vs 7.53 days; P = .32) or the mean length of hospital stay (11.1 vs 8.6 days; P = .063) between PTZ and carbapenem groups, respectively (Figure 3). Inpatient mortality was also comparable between the two groups (2.1% vs 0%; P = 1), with the majority of patients surviving at discharge.

Figure 3.

Secondary outcome.

Discussion

The results from this study suggest that PTZ could be used as an alternative to carbapenems for treatment of nonbacteremic UTIs due to ESBL-producing Enterobacteriaceae when isolates are susceptible based on lab testing. The successful use of PTZ for UTIs may partly be due to its ability to reach high concentrations in the urine. In a study conducted by Gavin et al., PTZ cured all patients with UTIs regardless of piperacillin MIC.¹² Although 100% cure was not demonstrated in the PTZ group in our study, the clinical response seen in our isolates with higher piperacillin MICs suggest that response of urinary isolates to PTZ may occur even when MIC is higher. Of note, two isolates in the PTZ group were reported to have a MIC of 32 μg/mL which at the time of treatment was reported as sensitive by the microbiology lab, though at the time of this study is now reported as intermediate. These isolates were included in the evaluation because they were reported to be sensitive at the time of treatment and reflect a realistic clinical situation where decision-making was based on sensitivity reports. Due to the perceived advantage of carbapenems and their status as preferred in the treatment of ESBL Enterobacteriaceae UTI, there was a larger proportion of patients receiving carbapenem therapy in the study time frame.

Our study does have some strengths to note. This study was conducted on data from multiple hospitals. While in the same general geographic area, the incorporation of multiple medical centers allowed us to evaluate data from urban, suburban, and rural patient populations. In addition to the incorporation of multiple centers and therefore a more varied patient population, we feel that there is opportunity for substantial practical application of the results as they represent data directly from clinical practice. We chose a composite endpoint of clinical response to focus on treatment outcomes rather than microbiologic outcomes. We feel that due to the complexities of treating patients with ESBL Enterobacteriaceae UTI, targeting an outcome of clinical response would contribute to the body of evidence regarding the use of PTZ in these clinical situations.

This study has several limitations that should be considered. First, the retrospective study design made it difficult to eliminate information and/or selection bias as well as any unknown confounding factors that may have affected the evaluation of treatment efficacy. We were also unable to ascertain the factors that affected physician decisions regarding choice of antimicrobial therapy, which was one of the factors used in determining clinical response. It is possible that the patients treated with PTZ requiring a change in antimicrobial therapy represented physician preference toward carbapenems once the ESBL isolate was realized, and these changes in therapy may not truly represent treatment failure (Table 3). Second, determining microbiological clearance was challenging since repeat urinalysis and urine cultures are not routinely recommended based on current guidelines. Third, assumptions regarding treatment efficacy were made due to the inability for patient follow-up after discharge. Fourth, the antimicrobial regimen at discharge was not standardized, and any deescalation to oral therapy that may have provided benefit is a potential confounder in our results. Fifth, classification of complicated and uncomplicated UTIs was not made as this differentiation relies on data that were not collected. Details concerning removal of catheters were an additional limitation in data collection. Finally, a true definition of UTI was not established in our study and was confirmed solely based on urine culture reports, physician documentation indicating clinical suspicion of UTI and decision to treat. It is likely that some urine culture reports represent colonization; however, this study was conducted at sites where antimicrobial stewardship programs have been implemented and treatment of asymptomatic bacteriuria is strongly discouraged. It was decided not to determine the presence of asymptomatic bacteriuria based on chart review alone as this may fail to capture urinary symptoms. Asymptomatic bacteriuria was documented in two patients, both in the carbapenem group. It was determined that this finding was minimal and did not require further evaluation. Inappropriate use of antimicrobials for treatment of asymptomatic bacteriuria is likely a common practice seen in various institutional settings.

Conclusions

In our retrospective cohort study, PTZ had a similar clinical response rate and showed no difference in outcome measures compared to carbapenems when used for treatment in patients with nonbacteremic UTIs due to ESBL-producing Enterobacteriaceae. While furthers studies are needed and carbapenems remain the preferred treatment in patients with a higher severity of illness such as those with bloodstream infections, our study indicates that PTZ may be an effective alternative in patients having ESBL-related infections solely from a urinary source.