Sepsis is the most common cause of death in

ADDRESSING ANTIMICROBIAL RESISTANCE IN THE INTENSIVE CARE UNIT * John P. Quinn, MD ABSTRACT Two of the more common strategies for optimizing antimicrobial therapy in the intensive care unit (ICU) are antibiotic cycling and de-escalation. Antibiotic cycling involves scheduled replacement of 1 or more antibiotics with another for a period of time, in order to avoid the development or control the rate of bacterial resistance. The data to support antibiotic cycling are sparse; its practice is supported more by theory, mostly among physicians in ICUs and pharmacists. Many of the positive results of published studies of cycling can be explained by differences in antibiotic usage or resistance levels at baseline. Also, recent data show an important and unintended consequence of cycling reducing resistance to 1 drug while developing resistance to its replacement. De-escalation involves an aggressive empiric therapy to cover the potential high minimum inhibitory concentration pathogens (eg, frequently Pseudomonas or Acinetobacter in the ICU). At 48 hours, reassessment helps to narrow down the therapeutic options to provide more specific antibiotic coverage, using a less toxic and often less expensive drug with a narrower spectrum. The data to support de-escalation are not widely available. Antibiotic cycling and de-escalation are both widely practiced in ICUs on an ad hoc basis. More careful clinical studies to assess these algorithms are under way. (Adv Stud Med. 2004;4(4A):S256-S261) *Based on a presentation given by Dr Quinn at a symposium held in conjunction with the 43rd Interscience Conference on Antimicrobial Agents and Chemotherapy. Professor, Division of Infectious Diseases, Rush University, Rush-Presbyterian St Luke s Medical Center, Chicago, Illinois. Address correspondence to: John P. Quinn, MD, 1900 West Polk, Room 1258, Chicago, IL 60612. E-mail: john_p_quinn@rush.edu. Sepsis is the most common cause of death in the intensive care unit (ICU). Infectious disease specialists are looking for novel strategies to control sepsis while practicing antimicrobial stewardship in light of the ever-increasing numbers of multiresistant bacteria. Two of the more common strategies for optimizing antimicrobial therapy in this setting are cycling and de-escalation. In order to understand the rationale for these strategies, it is important to understand the mechanisms of resistance. MECHANISMS OF RESISTANCE There are 4 determinants of the minimum inhibitory concentration (MIC) for beta-lactams in Gram-negative bacteria: penetration through the outer membrane, efflux, stability to beta-lactamases, and affinity for penicillin-binding proteins. In Gram-negative bacteria, efflux pumps interact with outer membrane channels (porins), forming multisubunit complexes that allow the bacteria to achieve multidrug resistance. Porins lie in the outer membrane of Gram-negative bacteria (Figure 1). 1 Through the use of accessory proteins, antibacterial agents are pumped out of the cell without ever having penetrated the inner membrane and thus gain access to penicillin-binding proteins. Porins are also able to block toxic compounds by themselves through several types of restriction: size, hydrophobicity, and charge. 1 Efflux pumps have gained much interest in the past decade as their structure and function are better understood, particularly in conjunction with porins and/or enzymatic hydrolysis. When considering antibiotic strategies, these mechanisms of resistance are important factors. ANTIBIOTIC CYCLING Antibiotic cycling involves scheduled replacement of 1 or more antibiotics with another for a period of time to avoid the development of bacterial resistance. The S256 Vol. 4 (4A) April 2004

main goal of cycling is to allow resistance rates for specific antibiotics to decrease, or at least remain stable. 2 The data to support antibiotic cycling are sparse; its practice is supported more by theory. While it may be embraced in some ICUs by physicians and pharmacists, molecular biology researchers tend to remain skeptical due to the difference in replication rates of most organisms (eg, every 20 minutes) compared with the cycling rates (eg, every 90 days). Nonetheless, some studies show a benefit with antibiotic cycling. The sentinel study in this area was published more than a decade ago. Gerding et al utilized aminoglycoside cycling in a hospital during a 10-year period. 3 A policy was instituted in which aminoglycosides were rotated based on the resistance pattern within the hospital. Table 1 shows the rotation schedule and declines in resistance during the study. 3 Of note, the reintroduction of gentamicin in study period 3 was rapid, as was the emergence of resistance to gentamicin. A more gradual introduction of gentamicin and removal of amikacin was used in period 5, without a corresponding rise in resistance rates. However, an important limitation of the study is the concomitant decreased usage of aminoglycosides in the hospital during the last 5 years of the study, due to advances in extended-spectrum betalactamase (ESBL) inhibitors and cephalosporins to kill Gram-negative bacteria. 3 The decrease in resistance may have been due in part to the decreased use of aminoglycosides in that environment. In 1997, Kollef and colleagues performed a cycling study in a cardiothoracic surgery unit. 4 A total of 680 patients undergoing cardiac surgery were evaluated for differences in bacteremia and ventilator-associated pneumonia (VAP) due to Gramnegative bacterial infections between two 6-month Figure 1. Schematic of Outer Membrane Porin and Efflux Pump in Gram-negative Bacteria OM = outer membrane. Adapted with permission from Nikaido H. Preventing drug access to targets: cell surface permeability barriers and active efflux in bacteria. Semin Cell Dev Biol. 2001;12(3):215-223. 1 Table 1. Resistance of Gram-negative Bacilli to Aminoglycosides During 5 Cycling Periods at the Minneapolis Veterans Affairs Medical Center (1980-1990) Resistant (%) Usage (%) Study Period Months (n) Isolates (n) Amikacin Gentamicin Tobramycin Amikacin Gentamicin Tobramycin 1 (baseline) 3 950 3.8 12.0 9.5 1.2 76.6 22.2 2 (amikacin) 26 6235 3.2 6.4* 4.8* 92.3* 5.3* 2.4* 3 (gentamicin) 12 2849 3.9 9.2* 6.0 31.5* 66.5* 2.1 4 (amikacin) 27 6115 3.1 5.8* 4.0* 97.5* 0.9* 1.6 5 (gentamicin) 51 12 333 2.9 5.7 4.2 29.8* 68.2* 2.0 *P <.001 compared with previous study period; P <.05 compared with previous study period. The observation period was divided into 5 distinct segments defined by predominant aminoglycoside use during that time. Amikacin and gentamicin went through 2 cycles of predominant use. Adapted with permission from the American Society for Microbiology. Gerding DN, Larson TA, Hughes RA, Weiler M, Shanholtzer C, Peterson LR. Aminoglycoside resistance and aminoglycoside usage: ten years of experience in one hospital. Antimicrob Agents Chemother. 1991;35(7):1284-1290. 3 Advanced Studies in Medicine S257

treatment periods: the before period, in which a third-generation cephalosporin (ceftazidime) was given, and the after period, in which a quinolone (ciprofloxacin) was given. Switching from ceftazidime to ciprofloxacin was associated with significant reductions in VAP (11.6% vs 6.7%; P =.028) and VAP attributed to Gram-negative bacteria (4.0% vs 0.9%; P =.013). The reduction in bacteremia due to Gram-negative bacteria was not significant (1.7% vs 0.3%; P =.125). Although the results suggest benefit with antibiotic switching, there was no difference in mortality rate from VAP due to Gram-negative bacteria, length of hospital stay, or development of multiorgan dysfunction between the 2 treatment periods. Importantly, the authors note that the results are preliminary and subject to several study limitations, including a small number of patients and lack of surveillance for any changes in colonization due to antibiotic class switching. 4 More recently, Kollef et al studied the effect of antibiotic cycling in a surgical ICU in a prospective study of 3668 surgical patients. 5 During each of 3 treatment periods, an antibiotic was chosen to treat Gram-negative bacterial infections (ceftazidime, Period 1; ciprofloxacin, Period 2; and cefepime, Period 3). The results showed a nonsignificant decrease in inadequate antimicrobial treatment for nosocomial infections but significant decreases in inadequate treatment for Gram-negative bacterial infections (Figure 2). 5 The mortality rate did not change among the 3 treatment periods, but the mean Acute Physiology and Chronic Health Evaluation (APACHE) II scores decreased across the treatment periods. In fact, the hospital mortality rate decreased significantly across treatment periods for the critically ill patients (APACHE 15), as shown in Figure 3. 5 Not surprisingly, the antibiotic resistance rates correlated with antibiotic use (ie, resistance rates for the particular study drug increased during its use in the study period; Table 2). 5 Rahal et al examined the effect of antibiotic restriction on the rates of resistance. 6 In a university hospital, a new antibiotic guideline excluded the use of cephalosporins without prior authorization by an infectious disease specialist, except for pediatric infection, single-dose surgical prophylaxis, acute bacterial meningitis, spontaneous bacterial peritonitis, and outpatient gonococcal infection. Imipenem also had the Figure 2. Inadequate Antimicrobial Treatment of All Nosocomial Infections According to Study Time Periods All nosocomial infections were microbiologically documented. Inadequate treatment was defined as the use of antibiotics with poor or no in vitro activity against the identified organisms causing infection, such as absence of antimicrobial agents directed at a specific class of microorganisms or administration of an antimicrobial agent to which the microorganism responsible for the infection was resistant. Inadequate treatment of Gram-negative and Gram-positive bacterial infections are also shown. Reproduced with permission from Kollef MH, Ward S, Sherman G, et al. Inadequate treatment of nosocomial infections is associated with certain empiric antibiotic choices. Crit Care Med. 2000;28(10):3456-3464. 5 Figure 3. Hospital Mortality According to Study Time Periods and APACHE Score APACHE = Acute Physiology and Chronic Health Evaluation. Reproduced with permission from Kollef MH, Ward S, Sherman G, et al. Inadequate treatment of nosocomial infections is associated with certain empiric antibiotic choices. Crit Care Med. 2000;28(10):3456-3464. 5 S258 Vol. 4 (4A) April 2004

same restriction except for use in medical ICU, surgical ICU, and cardiac ICU, with use limited to 72 hours. The rates of ceftazidime-resistant Klebsiella were compared between 1995 (the control period) and 1996 (the study period). With an 80% reduction in cephalosporin use during 1996 (72.5% reduction in ceftazidime use), the study results expectedly show reduced rates of ceftazidime-resistant Klebsiella. Unfortunately, there was a significant, concomitant increase in imipenem-resistant Pseudomonas aeruginosa from 1995 to 1996 (Table 3). 6 The investigators concluded that the class restriction significantly reduced nosocomial, plasmid-mediated, cephalosporin-resistant Klebsiella infection and colonization...at the price of increased imipenem-resistance in P aeruginosa...an overall reduction in multiply-resistant pathogens was achieved in 1 year. 6 However, one might argue that this is an example of shifting the sand on the beach from one hill to another. While not a treatment failure (and not strictly antibiotic cycling), it does serve as a useful example of the unintended consequences of shifting from one class to another. We recently completed a national surveillance study of 35 790 isolates of Gram-negative bacilli collected during 1994 through 2000 from more than 100 ICUs in 43 states plus the District of Columbia. 7 The study objective was to assess national rates of antimicrobial resistance among these isolates and compare them with rates of antimicrobial use. The results showed that amikacin and imipenem were the most active agents among all isolates and the 4 most common pathogens (Table 4) and that the activity of most agents decreased by 6% or less over the study period, with the exception of ciprofloxacin. 7 Among all isolates, the percentage of susceptibility to ciprofloxacin decreased by 10%; susceptibility decreased from 81% to 76% among P aeruginosa isolates. Importantly, ciprofloxacin resistance was associated with the 2.5-fold increase in fluoroquinolone use and cross-resistance to other broad-spectrum antimicrobial agents (Table 5). 7 As we noted, cross-resistance between fluoroquinolones and other classes has been associated with specific mechanisms of resistance (eg, ESBLs, plasmid-mediated resistance, increased efflux capability). 7 The next question is whether antibiotic cycling will affect these trends in resistance. The Centers for Disease Control and Prevention are sponsoring a 2-year cycling study at 3 US institutions. During the study, cefepime, imipenem, piperacillin/tazobactam, and ciprofloxacin will be rotated every 90 days, and outcomes, costs, and effect on flora (resistance and crossresistance patterns and mechanisms) will be measured. All clinical isolates will be saved, including weekly rectal swabs made before, during, and after antibiotic rotation, for organisms resistant to any of the 4 antibiotics under Table 2.Antibiotic Cycling and Resistance Patterns in a Surgical Intensive Care Unit Period 1 Period 2 Period 3 Ceftazidime Ciprofloxacin Cefepime Resistant Organism, n (%) (n = 1323) (n = 1243) (n = 1102) Gram-negative bacilli 35 (2.6) 4 (0.3)* 8 (0.7) resistant to thirdgeneration cephalosporin Gram-negative bacilli 2 (0.2) 14 (1.1)* 4 (0.4) resistant to ciprofloxacin Gram-negative bacilli 0 (0.0) 2 (0.2) 10 (0.9) resistant to cefepime *P <.05 comparing patients in Period 2 with Period 1; P <.05 comparing patients in Period 3 with Period 1; P <.05 comparing patients in Period 3 with Period 2 Adapted with permission from Kollef MH, Ward S, Sherman G, et al. Inadequate treatment of nosocomial infections is associated with certain empiric antibiotic choices. Crit Care Med. 2000;28(10):3456-3464. 5 Table 3. Source of Resistant Isolates in 1995 and 1996 During Cephalosporin Restriction Ceftazidime-Resistant Imipenem-Resistant Source of Klebsiella Pseudomonas aeruginosa patient isolates 1995 1996 1995 1996 Nosocomial 150 84 67 113 Nursing home acquired* 27 31 11 14 Outpatient 2 4 2 2 Emergency 3 1 1 3 Nursing home 10 16 2 8 Total 192 136 83 140 *A resistant isolate from a hospitalized patient within 72 hours after transfer from a nursing home. Isolates from patients in an affiliated nursing home, not admitted to the hospital. Reprinted with permission from Rahal JJ, Urban C, Horn D, et al. Class restriction of cephalosporin use to control total cephalosporin resistance in nosocomial Klebsiella. JAMA. 1998;280(14):1233-1237. 6 Advanced Studies in Medicine S259

rotation. To date, compliance is about 70% in the study. (Of note, in the published literature, compliance information is often lacking.) This study will allow us to understand the impact of antibiotic rotation on resistance selection (eg, the ability of cefepime to select ESBL-producing organisms) and the mechanisms of resistance (eg, independent mutations for cross-resistance). This study is under way and results are not yet available. ANTIBIOTIC DE-ESCALATION De-escalation is a new concept in antibiotic stewardship, so the data for this strategy are not widely available for review. Nonetheless, the concept is rather simple. Historically, treatment algorithms have employed a drug escalation strategy in which the most likely pathogen is surmised, based on the patient s level of sickness, use of steroids, presence of neutropenia, hospital setting, duration of hospitalization, and local resistance patterns. The most elegant antibiotic is chosen based on susceptibility spectrum, toxicity, and cost. At 48 hours, the patient is reassessed with culture results and clinical response. If the patient is responding, the regimen is continued. If not (ie, poor outcome or emergence of resistance), a more aggressive regimen is chosen. An important disadvantage with this approach is the loss of those first 48 hours, during which resistance can emerge or septic death can occur. Conversely, de-escalation involves an aggressive empiric therapy to cover the potential high-mic pathogens (eg, frequently Pseudomonas or Acinetobacter in the ICU). As with drug escalation, the clinician reassesses at 48 hours with culture results and clinical response. The culture results help to narrow the therapeutic options to provide more Table 4.Antimicrobial Susceptibility Rates for All and for the 4 Most Common Species of Gram-negative Bacilli, ICU Surveillance 1994-2000 All Isolates Absolute Pseudomonas Enterobacter Klebsiella Escherichia All Isolates, % Change in aeruginosa, % Species, % pneumoniae, % coli, % Antimicrobials (N = 35 790) % Susceptible (n = 8244) (n = 4999) (n = 4877) (n = 4027) Group 1, 90% susceptibility Amikacin 90 1 90 98 94 100 Group 2, 80%-90% susceptibility Imipenem 89 1 83 99 99 100 Tobramycin 83 3 87 92 85 96 Ciprofloxacin 81 10 76 90 88 97 Group 3 79% susceptibility Cefepime 78 2* 71 84 90 98 Ceftazidime 78 3 80 63 87 97 Gentamicin 78 6 68 92 85 95 Piperacillin/tazobactam 78 0 78 73 87 95 Aztreonam 71 4 67 68 86 97 Ticarcillin/clavulante 65 2 42 56 82 84 Piperacillin 59 4 74 60 39 64 Ceftriaxone 59 2 17 63 86 98 Cefotaxime 58 2 13 64 88 98 Ticarcillin 44 5 43 52 5 62 Ampicillin/sulbactam 35 3 2 21 65 63 *Based on % change, 1998-2000; Based on % change, 1997-2000. ICU = intensive care unit. Reprinted with permission from Neuhauser MM, Weinstein RA, Rydman R, Danziger LH, Karam G, Quinn JP. Antibiotic resistance among Gram-negative bacilli in US intensive care units. Implications for fluoroquinolone use. JAMA. 2003;289(7):885-888. 7 S260 Vol. 4 (4A) April 2004

Table 5. Examples of Antimicrobial Crossresistance Among Selected Gram-negative Bacilli, ICU Surveillance 1994-2000 Pseudomonas aeruginosa Enterobacter species Klebsiella pneumoniae Ciprofloxacin- Ciprofloxacin- Ciprofloxacin- Ciprofloxacin- Ciprofloxacin- Ciprofloxacin- Antimicrobial Resistant, % Susceptible, % Resistant, % Susceptible, % Resistant, % Susceptible, % Resistance (n = 1946) (n = 6298) (n = 486) (n = 4513) (n= 603) (n = 4274) Gentamicin 66.0 21.7 48.8 3.9 67.3 7.1 Ceftazidime 39.8 14.0 81.5 31.8 65.2 6.1 Imipenem 37.6 10.9 3.9 1.0 3.2 0.5 Amikacin 26.0 5.6 10.9 0.8 32.3 1.8 ICU = intensive care unit. Reprinted with permission from Neuhauser MM, Weinstein RA, Rydman R, Danziger LH, Karam G, Quinn JP. Antibiotic resistance among Gram-negative bacilli in US intensive care units. Implications for fluoroquinolone use. JAMA. 2003;289(7):885-888. 7 specific antibiotic coverage, using a less toxic and often less expensive drug with a narrower spectrum. 8 CONCLUSION Multiresistant organisms appear to arise not from a single genetic event but from a cascade of events. Using P aeruginosa as an example, initial use of cephalosporins may select derepressed type 1 betalactamase mutants or, less commonly, a plasmidmediated ESBL. Treatment with a quinolone will induce an antibiotic efflux pump. Treatment with a carbapenem such as imipenem will downregulate the D 2 porin, thus leading to multidrug resistance. Antibiotic cycling and de-escalation are both widely practiced in ICUs on an ad hoc basis. The current data to support these strategies are preliminary, but the strategies are in use nonetheless. As a result, more careful clinical studies to assess these algorithms are under way. REFERENCES 1. Nikaido H. Preventing drug access to targets: cell surface permeability barriers and active efflux in bacteria. Semin Cell Dev Biol. 2001;12(3):215-223. 2. Kollef MH. Is there a role for antibiotic cycling in the intensive care unit? Crit Care Med. 2001;29(4 suppl):n135-n142. 3. Gerding DN, Larson TA, Hughes RA, Weiler M, Shanholtzer C, Peterson LR. Aminoglycoside resistance and aminoglycoside usage: ten years of experience in one hospital. Antimicrob Agents Chemother. 1991;35(7):1284-1290. 4. Kollef MH, Vlasnik J, Sharpless L, Pasque C, Murphy D, Fraser V. Scheduled change of antibiotic classes: a strategy to decrease the incidence of ventilator-associated pneumonia. Am J Respir Crit Care Med. 1997;156(4 pt 1):1040-1048. 5. Kollef MH, Ward S, Sherman G, et al. Inadequate treatment of nosocomial infections is associated with certain empiric antibiotic choices. Crit Care Med. 2000; 28(10):3456-3464. 6. Rahal JJ, Urban C, Horn D, et al. Class restriction of cephalosporin use to control total cephalosporin resistance in nosocomial Klebsiella. JAMA. 1998;280(14):1233-1237. 7. Neuhauser MM, Weinstein RA, Rydman R, Danziger LH, Karam G, Quinn JP. Antibiotic resistance among Gram-negative bacilli in US intensive care units. Implications for fluoroquinolone use. JAMA. 2003;289(7):885-888. 8. Kollef MH. Optimizing antibiotic therapy in the intensive care unit setting. Crit Care. 2001;5(4):189-195. Advanced Studies in Medicine S261