UTILITY OF A COMBINATION ANTIBIOGRAM FOR TREATING PSEUDOMONAS AERUGINOSA

Empiric combination antibiotic therapy is often used to treat severe P. aeruginosa infections. Combination antibiograms have been employed to assist clinicians in selecting the most effective two-drug antibiotic regimens. The objectives of this study were to develop a combination antibiogram in order to compare betalactam monotherapy versus dual-therapy with a fluoroquinolone or aminoglycoside, identify optimal combination regimens, describe the differences between combination therapies with an aminoglycoside versus a fluoroquinolone and to evaluate susceptibility rates based on the source of infection and Intensive Care Unit (ICU) or non-ICU location. A retrospective observational study at a Veterans Affairs (VA) hospital in the Southwestern region of the U.S. was conducted. P. aeruginosa isolates were collected between January 2008 and February 2012 in hospitalized veterans. A total of 374 isolates were included, of which 61 (16%) were obtained from the ICU. Susceptibility rates for monotherapy with a beta-lactam ranged from 83.7 to 90.6%. Collectively, all P. aeruginosa isolates benefited in coverage with the addition of a fluoroquinolone or an aminoglycoside to one of the beta-lactams considered for monotherapy (p<0.01 for each comparison). Monotherapy with a beta-lactam could be considered for mild to moderate wound infections which had beta-lactam susceptibility rates greater than 90% and the addition of a fluoroquinolone did not significantly extend the spectrum. Combination susceptibility rates ranged from 89.0 to 99.2%. Dual therapy of a beta-lactam with amikacin or tobramycin resulted in significantly better coverage than with a fluoroquinolone (p<0.03 for all combinations). For severe infections dual therapy with tobramycin or amikacin may be preferred over fluoroquinolones, but the risks versus benefits of aminoglycoside therapy must be weighed for each patient. In conclusion, combination antibiograms are useful for evaluating the treatment of P. aeruginosa. Choosing the ideal antibiotic regimen ultimately deals with many factors and results of this combination antibiogram are only specific to this institution.


INTRODUCTION
Empiric combination antibiotic therapy is recommended for the treatment of P. aeruginosa infections particularly in the case of bacteremia, endocarditis, or pneumonia (Baddour et al., 2005;Baltch and Smith, 1985). The goal for using combination therapy to increase the likelihood of effective coverage, reduce the emergence of resistance and to provide additive or synergistic killing (Ibrahim et al., 2000;Iregui et al., 2002;Kollef et al., 1999;Garnacho-Montero et al., 2007;Lister and Wolter, 2005). Dual therapy usually consists of an antipseudomonal betalactam plus an aminoglycoside or fluoroquinolone, but increasing rates of fluoroquinolone resistance have decreased their efficacy in many areas. Lockhart et al. (2007) evaluated gram-negative bacilli resistance rates in ICUs from an average of 70 hospitals per year from 1993 to 2004. The percentage of P. aeruginosa isolates Science Publications AJID resistant to ciprofloxacin increased from an average of 11.2% at the beginning of the study period to 28.9% at the end of the study period.
In vitro studies have shown increased activity of either aminoglycosides or quinolones combined with a betalactam against P. aeruginosa, but there is limited clinical data to suggest that these combinations lead to improved patient outcomes (Giamarellou et al., 1984;Hollander et al., 1997;Karlowsky et al., 2005;Lister et al., 2006). Empiric combination therapy for P. aeruginosa bacteremia and ventilator-associated pneumonia has resulted in lower mortality compared to monotherapy (Hilf et al., 1989;Fowler et al., 2003). A meta-analysis by Paul et al. (2004) compared combination therapy with an aminoglycoside to monotherapy for severe infections and found that dual therapy did not provide a survival benefit and was associated with significantly greater rates of nephrotoxicity. Despite the inconclusive evidence for patient outcomes when using combination therapy, double coverage of P. aeruginosa remains a common practice.
To assist clinicians in selecting the most effective twodrug antibiotic regimen, combination antibiograms have been employed (Apisarnthanarak and Mundy, 2008;Christoff et al., 2010;Fox et al., 2008;Mizuta et al., 2006;Pogue et al., 2011). These antibiograms can be displayed in matrix form showing the percentage of isolates susceptible to various single agents and combinations of antibiotics. An advantage of a combination antibiogram is the ability to assess cross-resistance between specific antibiotics, whereas a standard antibiogram provides susceptibility rates for individual antibiotics. As susceptibility patterns are known to be institution-specific, it is important to develop antibiograms on a local basis. The aim of combination antibiograms is to permit clinicians to make a more informed decision in the section of empirical antimicrobial therapy, appropriately weighing the risks and benefit of combination antimicrobial treatment. With many areas having high rates of fluoroquinolone resistance it may be beneficial to evaluate combination susceptibility rates with fluoroquinolones and to compare it with aminoglycosides. Thus, the purpose of this study was to examine susceptibility rates for anti-pseudomonal beta-lactam agents, assessing the marginal effect of double-coverage with a fluoroquinolone or aminoglycoside. Differences in susceptibility rates between beta-lactam combinations and by isolate location and sources were also assessed.

Study Design and Patient Population
A retrospective observational study was performed at a single Department of Veterans Affairs (VA) facility located in the Southwestern region of the U.S. The study was approved by the local institutional review board prior to study initiation. A report was generated that included all positive cultures for P. aeruginosa isolates collected from patients with an inpatient hospital stay at the facility between January 2008 and February 2012. Detailed patient information was obtained through Computerized Patient Record System (CPRS). Patients had to be age 18 to 89 years at the time of culture collection and be hospitalized within seven days of the date when the isolate was obtained. For patients with multiple qualifying hospitalizations, only their first culture collection was considered. A total of 374 P. aeruginosa isolates from patients meeting these criteria were identified.

Data Collection and Study Definitions
Isolates were collected from patients in the Intensive Care Unit (ICU) and non-ICU locations. The Central Texas VA's main medical center is a 189 bed teaching hospital with 18 ICU beds. Isolates were cultured from various sources including the urine, sputum, wound, blood and other sources (i.e., bone, testes and ulna). Antimicrobial susceptibility of P. aeruginosa to antimicrobial agents commonly used for empirical coverage of gram-negative pathogens at this facility was determined. Optimal beta-lactam monotherapy of piperacillin/tazobactam, cefepime and imipenem/cilastatin was assessed and compared to combination therapy involving one of these antibiotics with a fluoroquinolone (ciprofloxacin or levofloxacin) or aminoglycoside (amikacin, gentamicin, or tobramycin). Each isolate was recorded as susceptible or resistant to a specific antibiotic by accessing CPRS for culture results. Intermediate susceptibilities were classified as resistant. For dual therapy susceptibilities, if the isolate was susceptible to at least one of the two antibiotics, then the isolate was classified as susceptible. This approach was modeled after several earlier studies utilizing combination antibiograms. The antibiotic sensitivities of P. aeruginosa were tested using the VITEK 2 automated system (bioMérieux) with one exception, P. aeruginosa isolates were tested for susceptibility to piperacillin/tazobactam by disk diffusion. All tests were performed in compliance with the current standards of the Clinical and Laboratory Standards Institute (CLSI).

Statistical Analysis
Monotherapy of the beta-lactam antibiotics versus double-coverage was compared collectively for all isolates, as well as by location and source. In order to examine improved susceptibility with combination therapy to beta-lactam monotherapy, McNemar's test

Science Publications
AJID was employed to test the null hypothesis of marginal homogeneity. Evidence supporting the alternative hypothesis would suggest that combination therapy was superior to antibiogram monotherapy. Proportion tests were also employed to compare dual therapies of betalactam with an aminoglycoside to beta-lactam with a fluoroquinolone. A type I error of α = 0.05 was assumed throughout. All analyses were performed using SAS, Version 9.2 (SAS Institute, Cary, NC).
Similar to the overall sample, patterns of increased coverage with dual therapy were observed among non-ICU isolates (p<0.01 for all comparisons). This was also true of most cases in the ICU, except for the addition of a fluoroquinolone to piperacillin/tazobactam, which did not significantly improve coverage.
Susceptibility of P. aeruginosa isolates from a urine source to monotherapy of a beta-lactam resembled that of the sample of all isolates ( Table 2). For isolates from a wound source, susceptibility rates were larger and less variable than that of the overall sample for monotherapy with each of the beta-lactam agents (ranging 90.2 to 91.8%). The greatest variability in susceptibility between beta-lactam agents was observed for isolates with blood sources 82.6% (cefepime and piperacillin/tazobactam) to 100% (imipenem/cilastatin). Coverage with monotherapy of a fluoroquinolone was reduced among urine sources (68.2% ciprofloxacin; 64.4% levofloxacin). Among isolates from urine and sputum sources, greater coverage was achieved with dual therapy, except for the case of imipenem/cilastatin + levofloxacin for sputum sources (91% versus 93.2%; p = 0.08). In many of the remaining sources, dual therapy resulted in 100% coverage. For wound and blood sources, dual therapy with a fluoroquinolone did not improve coverage. P. aeruginosa susceptibility rates were the lowest for monotherapy with a fluoroquinolone for blood isolates (69.6%). Amikacin resulted in 100% coverage for isolates from wound and blood sources. Table 1. Percentage of P. aeruginosa organisms susceptible to beta-lactam monotherapy versus combination antimicrobial therapy overall and by location (N = 374) Aminoglycoside Tobramycin  Ciprofloxacin  Levofloxacin  susceptible  susceptible  susceptible  susceptible  susceptible  Location Monotherapy    Levofloxacin  susceptible  susceptible  susceptible  susceptible  susceptible  Source Monotherapy   No significant differences in susceptibilities were found when comparing dual therapies involving ciprofloxacin versus dual therapies involving levofloxacin. Dual therapy of a beta-lactam with amikacin or tobramycin resulted in significantly greater coverage than with a fluoroquinolone (p<0.03 for all combinations; Table 3). Comparing locations, amikacin resulted in greater susceptibility in non-ICU settings when combined with any beta-lactam and ICU settings if combined with cefepime compared to a fluoroquinolone with a beta-lactam. Combination therapy with tobramycin was observed to have greater susceptibility than with a fluoroquinolone in the non-ICU settings, except for in the case of imipenem/cilastatin + levofloxacin (p = 0.07). Improved coverage with tobramycin versus other fluoroquinolones among isolates from the ICU was not detected. A difference in coverage between gentamicin and a fluoroquinolone with a betalactam was also not observed in the sample.

DISCUSSION
Several conclusions can be made based on the results of the combination antibiogram. Dual antibiotic therapy Science Publications AJID with a fluoroquinolone or an aminoglycoside was superior to monotherapy with a beta-lactam. Monotherapy with a beta-lactam could be considered for mild to moderate wound infections which had betalactam susceptibility rates greater than 90% and the addition of a fluoroquinolone did not significantly extend the spectrum. Ciprofloxacin and levofloxacin had very similar susceptibility profiles and were only 64 to 68% effective against urine cultures and did not provide significant benefit when combined with a betalactam for blood isolates. Imipenem/cilastatin outperformed other beta-lactams for blood and sputum isolates. Infections from the urine, sputum and blood had the most resistant isolates. For severe infections dual therapy with tobramycin or amikacin may be preferred over fluoroquinolones, but the risks versus benefits of aminoglycoside therapy must be weighed for each patient. Critically ill patients such as those with septic shock may be at greater risk for drug toxicity due to unstable pharmacokinetics and drug interactions. For approximately every 15 to 20 patients treated with dual therapy, 1 patient would benefit from tobramycin or amikacin versus a fluoroquinolone combination. If a P. aeruginosa isolate was resistant to a beta-lactam, amikacin would be active 87 to 92% of the time, tobramycin would be active 71 to 78% of the time and a fluoroquinolone would be active 29 to 43% of the time. Combination antibiotic therapy should be deescalated to a single active agent once microbiology and susceptibility results are confirmed. An interdisciplinary approach to managing aminoglycoside therapy including a pharmacy pharmacokinetic consult service may help to improve safety and optimize dosing and monitoring.
There are several limitations to interpreting the results of this study. It was observational and retrospective using isolates over several years and resistance rates fluctuate over time. We encountered a small sample size for patients in the ICU and for bacteremia. Therefore, this limits some of the conclusions we can make for these patients and increases our chances of making a Type II error. The results of this combination antibiogram are only specific to our institution. Resistance rates and frequencies can vary widely between institutions and even wards within the same institution. Another limitation to this study is that it did not examine clinical outcomes and future studies are needed to determine if susceptibility results are related to improved clinical outcomes. Although aztreonam typically has activity against P. aeruginosa, we did not include aztreonam in our susceptibility analysis because it is not routinely listed on our microbiology reports. Collecting susceptibility results for aztreonam would not have been feasible for this study. Similarly, colistimethate and polymyxin B were not included in this analysis.
Several studies have been published previously on the benefits of a combination antibiogram. A study by Mizuta et al. (2006) compared the susceptibility results of 3,134 P. aeruginosa isolates to determine optimal combinations for P. aeruginosa infections in hospitalized patients. They concluded the optimal combination therapy consisted of an aminoglycoside and a betalactam. Fluoroquinolones were not a component of the optimal combination for any year of the study. Another study by Fox et al. (2008) attempted to determine the in vitro efficacy of various dual antimicrobial combinations based on susceptibility results from 700 blood and respiratory isolates of Enterobacteriaceae and P. aeruginosa from hospitalized patients. They determined combination regimens with imipenem/cilastatin were the most effective. The addition of ciprofloxacin did not significantly improve the in vitro efficacy of piperacillin/tazobactam (87% versus 91%) or cefepime (89% versus 93%). A retrospective study by Christoff et al. (2010) looked at over 5000 isolates of gram-negative organisms, including P. aeruginosa, in the ICU. They reported combination antimicrobial therapy significantly increased the likelihood that the gram-negative organism was adequately covered in the ICU using piperacillintazobactam, ceftazidime, or imipenem/cilastatin plus an additional agent. They concluded that reporting antibiotic susceptibility data in the form of a combination antibiogram may be useful to clinicians who are considering empirical antimicrobial therapy in the intensive care unit. A study by Pogue et al. (2011) examined the susceptibility and resistance patterns of 467 gram negative ICU respiratory isolates. They found that the addition of a fluoroquinolone to an antipseudomonal beta-lactam did not substantially improve coverage, since nearly all beta-lactam resistant isolates were subsequently cross-resistant to fluoroquinolones. They concluded that optimal empiric therapy selection based on combination antibiogram data can identify the optimal combination regimen against all gram-negative ICU respiratory pathogens.

CONCLUSION
Combination antibiograms improve the ability to compare susceptibility rates with monotherapy versus Science Publications AJID dual-therapy, identify optimal combination regimens and quantify the difference between dual-therapy with a fluoroquinolone versus an aminoglycoside. Choosing the ideal antibiotic regimen ultimately deals with factors such as suspected pathogens, likely site of infection, patient allergies, drug penetration into different tissue sites and drug toxicity. Utilizing antibiograms to assist with the selection of empiric therapy is only one component of the decision-making process which may not apply to every hospitalized patient.

ACKNOWLEDGEMENT
This material is the result of work supported with resources and the use of facilities at the Central Texas Veterans Health Care System, Temple, Texas. Special thanks to physicians Dr. Chetan Jinadatha, Dr. Gabriella Kehler, Dr. David Allton and Dr. Gagan Prakash for reviewing the manuscript and providing helpful feedback.