Bone Marrow Transplantation (2002) 29, 367–371  2002 Nature Publishing Group All rights reserved 0268–3369/02 $25.00 www.nature.com/bmt Mini-review Antibiotic prophylaxis in patients receiving hematopoietic stem cell transplant

KA Sepkowitz

Clinical Infectious Diseases Section, Memorial Sloan-Kettering Cancer Center, New York, NY, USA

Summary: Oral prophylaxis: background and rationale

Effective prophylaxis against specific infections has allowed increasingly potent conditioning regimens to be For decades, various approaches have been tried in an given, thereby prolonging survival in HSCT recipients. attempt to reduce the risk of translocated oral and bowel The Centers for Disease Control and Prevention, in col- flora, which, along with central venous catheters, are the laboration with numerous professional societies, has source of the overwhelming majority of episodes of serious recently published guidelines to codify and advance this bacterial infection in the cancer and HSCT patient. The approach. Controversy remains in several areas but, goal is an important one and the intervention seems simple: curiously, the most intense debate concerns prevention just knock out the bowel and mouth flora with antibiotics. of bacterial infections, the most extensively studied of To this end, numerous regimens through the years have all of the approaches. Central to this debate are the been tried, included neomycin and polymyxin, trimethoprim– competing priorities of a potentially ill patient on the sulfamethoxazole and, most recently, the oral quinolones, one hand vs the long-term consequences of unchecked particularly ciprofloxacin.2–4 antibiotic use. The emergence in the 1990s of vancomy- Despite the logic of reducing bowel and mouth flora with cin-resistant Enterococcus demonstrated all too vividly antibiotics, thereby reducing the risk of subsequent bactere- how devastating such an end result could be. This arti- mia, there is no consensus that this is the right thing to do.5– cle will review the arguments for and against the routine 8 Preliminary work has even suggested that ciprofloxacin use of antibacterial prophylaxis in HSCT recipients. without metronidazole given as prophylaxis might promote Bone Marrow Transplantation (2002) 29, 367–371. DOI: development of graft-versus-host disease9 and that cipro- 10.1038/sj/bmt/1703366 floxacin alone may be associated with an increased risk of Keywords: bacterial infections; prophylaxis; drug resist- leukemia relapse among HSCT recipients.10 ance; sepsis In general, the infectious disease community, citing the concern of promoting drug resistance, has been hesitant about making this a routine approach, while the oncology community, with a specific patient in front of them whom Clinicians have long sought a means of preventing infec- they are trying to keep out of the hospital, has been more tions among those receiving hematopoietic stem cell trans- enthusiastic. plantation (HSCT). Although advances have been made in What then is the evidence for routine use of oral antibac- preventing certain infections such as Pneumocystis carinii terial prophylaxis? A well-conducted meta-analysis of trials 1 and Candida albicans, prevention of bacterial infections using ciprofloxacin as the agent of choice for prophylaxis arising from routine enteric flora, which contribute a sub- of patients with neutropenia was published in 1996.2 The stantial amount of infectious morbidity, has been more dif- authors compared trials of various quinolones vs another ficult. prophylactic treatment (trimethoprim–sulfamethoxazole, This review will consider both sides of the ongoing argu- non-absorbable agents, or placebo). In all, results from ment on optimal management: the noble goal of briefer and 2112 patients were included, representing 19 studies. fewer hospitalizations with less mortality vs the inevitable The authors found that, as hoped, the rate of Gram-nega- consequence of antibiotic use: drug resistance. tive bacteremias was reduced. Furthermore, unlike many other reports, there was no evidence for an increase in Gram-positive bacteremias. Both of these findings are exciting and important. Furthermore, in general, most clin- icians believe their patients benefit from therapy. That was the good news. The more problematic finding Correspondence: KA Sepkowitz, Infectious Diseases Service, Memorial was that neither days with fever nor patient mortality was Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY reduced. In other words, the intervention successfully low- 10021, USA ers risk for certain types of bacteremia, some of them Antibiotic prophylaxis in HSCT recipients KA Sepkowitz 368 extremely serious, but, in the final analysis, has very little tuberculosis strains resistant to streptomycin. A similar positive impact on the patient’s overall health or survival. phenomenon occurred when penicillin was introduced into A more recent meta-analysis examining a comparable hospitals in the 1950s: Staphylococcus aureus quickly group of studies reached a similar and perhaps even more developed resistance and rendered the drug useless in this cautious conclusion.3 In this report, which included 1408 therapeutic context. And so within moments after the wide- subjects from 18 studies (some included in the first meta- spread introduction into practice of that great twentieth cen- analysis), the authors found a reduction in infection-related tury miracle – antibiotics – the bitter irony of drug resist- outcomes but no difference in mortality. Furthermore, the ance was born. difference in fever-days was modest overall and non-exi- The two basic tenets governing this unfortunate circum- stent in blinded studies. stance are simple and unalterably true: (1) the bacteria, Not only that, the authors voiced a concern about the sooner or later, will always prevail, given their vast num- long-term consequences of such an approach: not necessar- bers; (2) the more antibiotic given, the faster resistance ily to the actual patient at hand, but to future patients. emerges. Specifically they raised the specter of drug resistance The fate of Escherichia coli susceptibility is quite telling resulting from unchecked antibiotic use. As they concluded, in this regard11 (Table 2). E. coli is the most common the emergence of resistant bacteria ‘threatens to undermine Gram-negative bacterium causing bacteremia among neu- the long-term efficacy of prophylaxis with fluoroquino- tropenic hosts in most series. Researchers from the EORTC lones’. followed the fate of E. coli as quinolone use was introduced Resistant bacteria are only one of several real concerns into common practice in Europe. They identified two arising from the practice of routinely prescribing oral important consequences. First, the prevalence of resistant agents to prevent bacterial infections. Also important to E. coli rose quickly during the 11-year study period (from consider are other unintended consequences of this none to 28% resistance), which was hardly a surprise. approach, each of them with potentially deleterious effects Second, and more unexpectedly, the rate and resistance pat- not just to the current group of patients being treated with tern to the quinolones of coagulase-negative Staphylo- HSCT, but those who will require this treatment in the com- coccus increased sharply as well, from none to 61% resist- ing years (Table 1). Although the practice of oral prophy- ance. This demonstrated that, in addition to the desired (and laxis is all but routine in many HSCT centers, the current undesired) effect of an intervention, any change in approach problems with drug resistance may force a careful recon- may exert an unintended influence, in this case on an ‘inno- sideration of this still unproved approach. cent bystander’ pathogen that has emerged as the most Clinical situations with clear risk from a predictable and common cause of catheter-related bacteremia. This is dis- easily prevented organism, such as Streptococcus pneumon- cussed in more detail below under ‘unexpected pathogens’. iae in patients with chronic GVHD, lie outside the debate.1 In this example, the consequences of no intervention far outweigh those arising from chronic penicillin therapy and Vancomycin-resistant Enterococcus prophylaxis must be given.1 The E. coli data alarmed some but not all practitioners. For many, the threat of emergent drug resistance, similar to the Emergence of drug resistance threat of military bioterrorism, was considered the result of the overheated imagination of the infectious Gram-negative organisms disease/infection control community. Yet this dark fear became everyone’s collective nightmare in the 1990s with Within months of the first doses of streptomycin, given for the rapid appearance internationally of vancomycin-resist- tuberculous meningitis in the 1940s, came reports of ant Enterococcus (VRE).12 Few hospital-based infections

Table 1 Arguments for and against use of routine prophylaxis to prevent bacterial infections among HSCT recipients

Arguments in favor Arguments against

Gut decontamination Fewer Gram-negative bacteremias No change in mortality rates Possibly less fever Possibly more Gram-positive infections Possibly briefer hospitalization Loss of quinolones as subsequent therapeutic option Promotion of antibiotic resistance Promotion of non-susceptible organisms (eg, S. viridans) Monotherapy Less toxicity Possibly blunts emergence of resistance Less nurse/pharmacy time committed Possibly more efficient bacterial killing Broader initial empiric spectrum Out-patient antibiotics Patient comfort and quality of life Practicality Lower cost Safety Possibly lower risk of nosocomial complications

Bone Marrow Transplantation Antibiotic prophylaxis in HSCT recipients KA Sepkowitz 369 Table 2 Use of fluoroquinolone prophylaxis and frequency of fluoroquinolone-resistant bacteremic episodes in neutropenic patients with cancer

EORTC EORTC EORTC MSKCC 1983–1985 1986–1990 1991–1993 2000

Prophylaxis given 3/214 (1.4%) 228/694 (33%) 318/706 (45%) NA Resistance 0/26 0/66 11/40 (28%) 14% E. coli Coagulase-negative 0/22 44/172 (26%) 23/38 (61%) 62% Staphylococcus P. aeruginosa 1/25 (4%) 1/39 (3%) 1/13 (8%) 17% K. pneumoniae 0/3 1/17 (6%) 1/13 (8%) 3%

EORTC and Memorial Sloan-Kettering Cancer Center (MSKCC) data (adapted from Ref. 10 and unpublished data). Quinolone prophylaxis is not routinely used at MSKCC. Data demonstrates relationship between increasing quinolone use and rising rates of drug resistance. MSKCC data, although collected 7 years after EORTC data, reflects relative persistence of quinolone susceptibility, possibly related to less frequent quinolone use in prophy- laxis context. have had a more profound impact on transplant and infec- intended bacteria; it is quite another to somehow fundamen- tion control units than this pathogen. It has proven difficult tally alter which bacteria dominate in the cancer setting. A to treat and, even with more and more effective agents, vivid example of this is S. viridans, which emerged in many VRE remains a cause for tremendous alarm when it appears cancer centers soon after the introduction of widespread use on any in-patient ward. of quinolone prophylaxis.16–19 Such a result might have How did this epidemic happen? Although numerous been anticipated, given the prevalence of S. viridans in nor- studies have yielded different explanations, inevitably the mal oral mucosa and the relative inactivity of the quinolone root cause is overuse and misuse of vancomycin. Sub- class, including the newer agents, against this pathogen. sequent spread via traditional nosocomial pathways then Several reports from the early 1990s described the dramatically amplified its presence. emergence of S. viridans. At MD Anderson, the rate of S. VRE now stands as the very ‘poster child’ of antibiotic viridans bacteremia rose from 1/10 000 to 47/10 000 overuse, yet its devastating presence has taught very little admissions.16 The increase was associated with use as in terms of preventing the next nosocomial outbreak. The prophylaxis of either trimethoprim–sulfamethoxazole or next bacterium might be a vancomycin-resistant strain of ciprofloxacin. The investigators also described a shock-like Staphylococcus,aPseudomonas aeruginosa resistant to all syndrome from the organism, which previously had seldom antibiotics but polymixin, or yet another pathogen, but it is been encountered.16 The disease resembled toxic shock and sure to have comparable consequences. was characterized by hypotension, rash, palmar desqua- As important as the demonstration that bad things can mation, and adult respiratory distress syndrome. The infec- happen to good hospitals, the emergence of VRE also tion was fatal in 26% of patients. Subsequent reports from showed the fundamental folly in the premise that somehow other groups have verified that S. viridans can indeed cause scientists can forever keep a step ahead of the bacteria. a shock-like illness. In addition, use of ara-C (cytosine Infectious disease specialists have little or no faith in arabinoside)17 or cyclophosphamide18 has been shown to be science in this race and oncologists should become equally associated with development of S. viridans, presumably due pessimistic. As always, no sooner has a ‘wonder drug’ such to the increased intensity of oral mucositis associated with as linezolid been approved than the inevitable converse of these chemotherapeutic agents. antibiotic potency, drug resistance, is reported.13,14 Thus the short-term gain afforded by quinolone use was Thus, empiric use of vancomycin should be sharply in danger of being overtaken by the delayed consequence restricted to those few centers with a large problem with of the approach: a new organism that could cause shock Streptococcus viridans or other agents that may cause rapid with high mortality rates and, by coincidence, was itself death.1,15 Widespread, empiric use would almost certainly becoming increasingly resistant to such first-line agents cause more harm than it would ameliorate: one pathogen as penicillin.20 (S. viridans) would simply be traded for the next (VRE). A related phenomenon is the emergence of environmen- Importantly, the relatively low virulence of coagulase-nega- tal organisms of modest pathogenic capacity, such as Steno- tive Staphylococcus allows clinicians some time to establish trophomonas maltophilia21 and Acinetobacter species.22 the optimal antibiotic regimen with minimal adverse These organisms have established a foothold in various clinical consequence. intensive care units because they are resistant to the com- mon potent antibiotics used for the very ill, such as imip- Unexpected pathogens enem or the third-generation cephalosporins. The result of this type of antibiotic overuse is infestation to the point of As mentioned above, an unanticipated consequence of anti- endemicity in numerous ICUs of tenacious, drug-resistant biotic use is the emergence of unexpected pathogens. It is bacteria that are a particular risk to the already compro- one thing to influence the drug susceptibility profile of the mised, such as those receiving HSCT. Once they establish

Bone Marrow Transplantation Antibiotic prophylaxis in HSCT recipients KA Sepkowitz 370 their ecosystem in an ICU, it is very difficult to get them This is particularly problematic among cancer and trans- out, since they thrive in tepid water, respirator tubing, as plant patients. These individuals are hospitalized repeatedly well as on inanimate surfaces, such as tabletops and bed and for protracted periods, often requiring prolonged guardrails. Prevention of these bacteria is far easier than courses of multiple antibiotics. Drug allergy and intoler- treatment, underlining again the need for a disciplined pro- ability are at least as common in the cancer or transplant gram of antibiotic control and will-power against the patient as the general population, in whom up to 25% report seductions of immediate broad-spectrum activity. Aware- allergies to at least one agent.24 Thus, for the beta-lactam ness of potential long-term consequences should inform all allergic patient, removal of the quinolone class as a poten- antibiotic decision-making. tial option relegates choices to the more toxic, such as the A final example of the dangers of disturbing normal host aminoglycoside class and/or an agent such as vancomycin, flora, unless necessary, was demonstrated in a study that which may itself promote drug resistance. Preserving the examined VRE density as a function of anti-anaerobic anti- option of quinolone use is a worthy goal for management biotic use.23 Investigators found that anti-anaerobic anti- of such patients. biotics appeared to promote high-density VRE colonization that in turn was associated with a higher likelihood of VRE transmission. Thus, antibiotics used to prevent a first infec- Summary and recommendations tion may inadvertently lead to development of a second and even more serious condition. Bacterial infections, although perhaps less dramatic or The optimal approach to prevention and control of VRE intellectually stimulating as their viral and fungal counter- continues to develop. New investigational non-absorbed parts, continue to account for significant morbidity and antibiotics such as ramoplanin offer one potential approach, mortality in HSCT recipients. Furthermore, the possibility while other centers routinely screen all HSCT pre-trans- of preventing these complications seems uncharacteristi- plant and on re-admission to identify VRE carriers. One cally simple. The common pathogens are known; numerous center both screens and quantifies the amount of VRE to oral agents are active against them, and patients usually can better determine patient risk for invasive disease, as well tolerate the medications. What then is the problem? as the influence of various antibiotics on VRE carriage (R As discussed above, the long-term consequences of rou- Bonomo and M Lisgaris, personal communication). tine bacterial prophylaxis are beginning to emerge and are causing significant concern in the infectious disease com- munity. Rising rates of resistance, emergence of unexpected Subsequent antibiotic choices pathogens, and loss of quinolones in the treatment of fever and neutropenia all argue strongly for containment of rou- In addition to promotion of drug resistance and a distortion tine use. That said, there remains a significant need for such of the hospital spectrum of anticipated agents towards more agents in certain situations. Patient comfort and confidence tenacious and perhaps more virulent replacements, the are extremely important issues, particularly over the long indiscriminate use of the quinolones has a potentially del- haul of cancer chemotherapy or HSCT. eterious effect on subsequent antibiotic choices. Patients One approach would be to use oral prophylaxis cau- receiving quinolones as prophylaxis who then develop neu- tiously, restricting its use to the highest risk patients (Table tropenia and fever (and fever episodes in most studies are 3). These include persons with previous serious Gram- not decreased by quinolone prophylaxis) must be hospi- negative infection, such as ecthyma gangrenosum from talized for intravenous antibiotics. However, these patients Pseudomonas aeruginosa. In addition, those with known should not receive quinolones in management of their neu- predictable immune defects, such as the propensity for per- tropenic fever on the assumption that the infection has sons with chronic GVHD to develop invasive Streptococcus ‘broken through’ the prophylaxis. Thus, an entire class of pneumoniae, should receive prophylaxis. Prophylaxis might agents is removed from the treating team’s armamentarium. also be carefully considered for older patients and those

Table 3 Recommendations for use of bacterial prophylaxis in HSCT

Agent Indication

Trimethoprim–sulfamethoxazole Restrict use: limited indication for transient gut decontamination Amoxicillin-clavulanic acid (Augmentin) Restrict use: limited indication due to narrow spectrum against high-risk bacteria Quinolone Restrict to patients with recent history of severe gram-negative infection or significant co- morbidities Vancomycin Restrict to few centers with high or rising rates of severe, penicillin-resistant S. viridans Ramoplanin Utility unknown: in studies to determine if non-absorbable agents with narrow spectrum might prevent VRE Penicillin Prevention of S. pneumoniae infections in persons with chronic GVHD. Unclear if rising S. pneumoniae resistance rates impel use of alternative agent, such as levofloxacin

Bone Marrow Transplantation Antibiotic prophylaxis in HSCT recipients KA Sepkowitz 371 with significant co-morbidities. Although this parsimonious 11 Cometta A, Calandra T, Bille J, Glauser MP. Escherichia coli approach flies in the face of current practice and will be resistant to fluoroquinolones in patients with cancer and neu- uncomfortable in the near term, the ultimate consequences tropenia. New Engl J Med 1994; 330: 1240–1241. of using less oral prophylaxis will benefit current and future 12 Low DE, Keller N, Barth A, Jones RN. Clinical prevalence, HSCT recipients. antimicrobial susceptibility, and geographic resistance patterns of enterococci: results from the SENTRY Antimicrobial Sur- veillance Program, 1997–1999. Clin Infect Dis 2001; 32 References (Suppl. 2) S133–S145. 13 Tsiodras S, Gold HS, Sakoulas G et al. Linezolid resistance 1 Centers for Disease Control and Surveillance. Guidelines for in a clinical isolate of Staphylococcus aureus. Lancet 2001; preventing opportunistic infections among hematopoietic stem 358: 207–208. cell transplant recipients: recommendations of the CDC, the 14 Gonzales RD, Schreckenberger PC, Graham MB et al. Infec- Infectious Disease Society of America, and the American tions due to vancomycin-resistant Enterococcus faecium Society of Blood and Marrow transplantation. MMWR Morb resistant to linezolid. Lancet 2001; 357: 1179. Mortal Wkly Rev 2000; 49:1–125. 15 Attal M, Schlaifer D, Rubie H et al. Prevention of gram- 2 Cruciani M, Rampazzo R, Malena M et al. Prophylaxis with positive infections after bone marrow transplantation by sys- fluoroquinolones for bacterial infections in neutropenic temic vancomycin: a prospective, randomized trial. J Clin patients: a meta-analysis. Clin Infect Dis 1996; 23: 795–805. Oncol 1991; 9: 865–870. 3 Engels EA, Lau J, Barza M. Efficacy of quinolone prophylaxis 16 Elting LS, Bodey GP, Keefe BH. Septicemia and shock syn- in neutropenic cancer patients: a meta-analysis. J Clin Oncol drome due to viridans streptococci: a case-control study of 1998; 16: 1179–1187. predisposing factors. Clin Infect Dis 1992; 14: 1201–1207. 4 Nagatomo A, Watanabe K, Kunikane H et al. A randomized 17 Bochud PY, Eggiman P, Calandra T et al. Bacteremia due controlled trial of sulfamethoxazole/trimethoprim plus nor- to viridans streptococcus in neutropenic patients with cancer: floxacin versus sulfamethoxazole/trimethoprim alone for the clinical spectrum and risk factors. Clin Infect Dis 1994; 18: prophylaxis of bacteria infection during chemotherapy for 25–31. lung cancer. Lung Cancer 1998; 19: 121–125. 18 Marron A, Carratala J, Gonzalez-Barca E et al. Serious com- 5 Murphy M, Brown AE, Sepkowitz KA et al. Fluoroquinolone plications of bacteremia caused by Viridans streptococci in prophylaxis for the prevention of bacterial infections in neutropenic patients with cancer. Clin Infect Dis 2000; 31: patients with cancer. Clin Infect Dis 1997; 25: 346–347. 1126–1130. 6 De Pauw BE, Donnelly JP. Infections in the immunocompro- 19 Bochud PY, Calandra T, Francioli P. Bacteremia due to virid- mised host: general principles. In: Mandell GL, Bennett JE, ans streptococci in neutropenic patients: a review. Am J Med Dolin R (eds). Principles and Practice of Infectious Disease, 1994; 97: 256–264. 5th edn. Churchill-Livingstone: Philadelphia, PA, 2000, pp 20 Diekema DJ, Beach ML, Pfaller MA, Jones RN. Antimicrobial 3079–3090. resistance in viridans group streptococci among patients with 7 Baum HV, Franz U, Geiss HK. Prevalence of ciprofloxacin- and without the diagnosis of cancer in the USA, Canada and resistant Escherichia coli in hematologic–oncologic patients. Latin America. Clin Microbiol Infect 2001; 7: 152–157. Infection 2000; 28: 278–281. 21 Alfieri N, Ramotar K, Armstrong P et al. Two consecutive 8 Yeh SP, Hsueh EJ, Yu MS et al. Oral ciprofloxacin as antibac- outbreaks of Stenotrophomonas maltophilia (Xanthomonas terial prophylaxis after allogeneic bone marrow transplan- maltophilia) in an intensive-care unit defined by restriction tation: a reappraisal. Bone Marrow Transplant 1999; 24: fragment-length polymorphism typing. Infect Control Hosp 1207–1211. Epidemiol 1999; 20: 553–556. 9 Beelen DW, Elmaagacli A, Muller KD et al.Influence of 22 Bergogne-Berezin E. The increasing role of Acinetobacter intestinal bacterial decontamination using metronidazole and species as nosocomial pathogens. Curr Infect Dis Rep 2001; ciprofloxacin or ciprofloxacin alone on the development of 5: 440–444. acute graft-versus-host disease after marrow transplantation in 23 Donskey CJ, Chowdhry TK, Hecker MT et al. Effect of anti- patients with hematologic malignancies: final results and long- biotic therapy on the density of vancomycin-resistant enteroc- term follow-up of an open-label prospective randomized trial. occi in the stool of colonized patients. New Engl J Med 2000; Blood 1999; 93: 3267–3275. 343: 1925–1932. 10 Carlens S, Ringden O, Aschan J et al. Risk factors in bone 24 Lee CE, Zembower TR, Fotis MA et al. The incidence of marrow transplant recipients with leukaemia. Increased antimicrobial allergies in hospitalized patients: implications relapse risk in patients treated with ciprofloxacin for gut regarding prescribing patterns and emerging bacterial resist- decontamination. Clin Transplant 1998; 12:84–92. ance. Arch Intern Med 2000; 160: 2819–2822.

Bone Marrow Transplantation