Altclas and colleagues and Drs. Castagnola and Viscoli have cited many of the disadvantages of selective quinolone prophylaxis, including colonization and subsequent infection by opportunistic yeasts; reduction in prophylactic efficacy associated with the increasing prevalence of quinolone-resistant microorganisms in many centers; and lack of consistent effect on such outcomes as febrile morbidity, infection-related mortality, and the pattern of prescribing parenteral antimicrobial agents. Despite these criticisms, a reduction in the incidence of infection with aerobic gram-negative bacilli has been the outcome most consistently shown in clinical trials. However, differences in mortality rates related to gram-negative infection have been harder to demonstrate because of the confounding bias of the highly effective parenteral regimens of antibacterial agents for gram-negative organisms now available. Limited experience suggests that successful prophylaxis can reduce the need for parenteral antibacterial therapy [1, 2]; however, larger, carefully designed, randomized, controlled clinical trials are required for validation.

Some investigators have interpreted the lack of effect on febrile morbidity as evidence of a transformation from microbiologically documented infectious episodes into unexplained fevers. Altclas and colleagues reject the provocative alternate hypothesis that some unexplained fevers may represent a process other than invasive bacterial infection. The pathogenesis may be related to the absorption of pyrogenic substances through intestinal epithelial surfaces damaged by cytotoxic therapy, an effect analogous to that seen with permeability probes such as lactulose [3]. The clinical demonstration that antibacterial therapy can be safely withheld or discontinued early in neutropenic patients with unexplained fevers would support this hypothesis [1, 2, 4]. Because unexplained fevers represent approximately 40% of febrile neutropenic episodes, a better understanding of the pathogenesis would improve our ability to identify patients most likely to benefit from antibacterial therapy.

Although it may be true that the disadvantages may offset the benefits accrued from a reduction in the rate of gram-negative and gram-positive infections, these arguments may not be relevant in all centers. Prophylaxis policies may be tailored for specific circumstances. For example, yeast colonization among patients receiving ofloxacin plus rifampin may not be an issue in centers where fluconazole antifungal prophylaxis is the policy. The prevalence of quinolone-resistant aerobic gram-negative bacilli may be low in centers that have restrictive policies for prescribing antimicrobial agents. We agree with Drs. Castagnola and Viscoli that quinolone-based prophylaxis should be restricted to patients in whom benefit has been demonstrated, including patients receiving remission-induction therapy for acute myelogenous leukemia or those undergoing hematopoetic stem cell transplantation.

Our experience suggests that selective quinolone-based prophylaxis strategies can be useful but are not applicable in all circumstances or in all centers. Accordingly, generalized statements for or against the use of these strategies still seem inappropriate.