1. ¹Brigham and Women's Hospital, Dana-Farber Cancer Institute, Boston, MA, USA
2. ²Harvard Medical School, Boston, MA, USA

Correspondence: Dr LR Baden, Brigham and Women's Hospital, PBB-A4, 15 Francis Street, Boston, MA 02115, USA. Fax: +1 617 732 6829

Received 20 January 2003; Accepted 19 February 2003.

Prevention and treatment of invasive fungal infections (IFI) in bone marrow transplantation (BMT) are essential to successful outcome. Unfortunately, preventative therapies for fungal infections are often toxic and fungal infections are usually detected at an advanced stage, thus increasing the challenge for successful treatment. Critical issues in managing these infections are identifying them as early as possible and treating them aggressively with a multimodality approach (eg augmentation of host immunity, utilization of growth factors, control of permissive viral copathogens, and antimicrobial therapy). There are several important advances in this field, which will be discussed: (1) approach to prophylaxis in BMT patients, (2) antifungal susceptibility testing, (3) novel diagnostic strategies, and (4) new therapeutic approaches.

Understanding who is at risk and when the risk occurs is critical. With this in mind, there are three principles that are important to consider in the approach to fungal disease in this patient population. The first is pathogen exposure: colonization precedes disease. Colonization with Candida sp. is common and typically does not elicit any overt clinical response nor does it demand a therapeutic intervention, as Candida sp. may be part of the normal gastrointestinal and/or vaginal flora. Aspergillus colonization is extremely worrisome in the neutropenic patient and almost always requires an intervention. The second principle is host susceptibility, or the net state of immunosuppression. In the allogeneic vs autologous BMT patient, the duration of neutropenia, the development and treatment of graft-versus-host disease (GvHD), the administration of growth factors, and the presence of immunomodulating viruses (cytomegalovirus CMV, HHV-6) all have serious repercussions on the immune system. It is difficult to calculate immune impairment, so the timing and deployment of antimicrobials will be different for each patient. Previously, prolonged neutropenia imparted the greatest risk for IFI. Now the greatest risk in BMT patients may not be during neutropenia, but months later associated with the occurrence of GvHD. These patients need a modified pre-emptive approach to infection.¹ Another issue to consider are the effects of novel immunotherapies (eg new conditioning regimens, monoclonal antibody therapy) (Marty FM et al. 42nd ICAAC, 2002; Abstract M-1234). As the implications of these new therapies emerge, we will have to modulate our preventative approach accordingly. A major challenge in developing diagnostics for detecting and therapeutics for treating IFIs is the limitations of the current gold standard case definition: a consensus panel defining proven, probable, and possible IFI based on clinical, radiographic, microbiologic, and histopathologic criteria.² Recognizing the reservoir of patients colonized with epidemiologically important pathogens and proper identification of the period(s) of increased host susceptibility, will facilitate the development of rational preventative strategies for IFI.

There are three strategies to prevent IFI. The first is prophyl-axis, where a therapy is given to all patients at risk: for example, fluconazole given to all patients undergoing an allogeneic BMT to prevent Candida infection. The second is empiric therapy, where medication is given to patients with a presumed infection: for example, amphotericin B use in the setting of febrile neutropenia refractory to antibacterial antimicrobials agents. The final approach is pre-emptive, where a radiographic or molecular marker suggesting the presence of an IFI would trigger antifungal therapy. A currently deployed pre-emptive strategy is ganciclovir use when CMV replication is detected in the blood stream post-BMT but prior to defined tissue infection. Each of these approaches will be addressed in turn.

There are two randomized controlled trials demonstrating a benefit of fluconazole prophylaxis in the allogeneic BMT setting. In these two publications, fluconazole was administered at a dose of 400 mg/day during the transplant period until engraftment³ or until day 75,⁴ both studies demonstrated a decrease in the incidence of Candida sp. infections. A prolonged survival benefit was reported in one.⁵ A consequence of widespread fluconazole use, not measured in these two reports, is the effect of antimicrobial pressure on the selection for and emergence of fluconazole-resistant organisms, such as Candida kruseii and glabrata.⁶ Subsequently, multiple studies have demonstrated this occurrence. In one report, Wingard et al⁷ demonstrated a 40% colonization rate and a seven-fold increase in C. kruseii infections in patients on fluconazole prophylaxis. A study conducted in Seattle from 1994 to 1997, in which patients who underwent allogeneic BMT received fluconazole prophylaxis, demonstrated increased colonization with more resistant Candida sp. (38% with C. kruseii and 37% with C. glabrata) and colonization was associated with the development of fungemia with the colonizing organism.⁸ In the short time frame fluconazole has been used as prophylaxis in the allogeneic BMT setting, there has been a shift in the species of yeast causing invasive disease – from C. albicans/tropicalis to C. glabrata/kruseii. It is important to note that the overall rate of fungal infection has decreased, but the more resistant organisms have increased. An additional finding in the Seattle study was that the two fungemias with C. albicans that occurred were noted to be fluconazole resistant. A clear benefit of azole prophylaxis in patients undergoing chemotherapy for acute leukemia has not been demonstrated,⁹ although some authors advocate their use in this setting.¹⁰

In treating these patients, the consequence of the emergence of resistant microbes needs to be carefully considered.^11,12,13,14 If a patient is on an azole for a period of time and then develops an IFI, it is unlikely that the offending pathogen will be susceptible to the currently used azole,¹⁵ even though the speciation may suggest that the organism may be (C. albicans/parapsilosis/tropicalis are typically susceptible, while C. glabrata is dose responsive, and C. kruseii is intrinsically resistant to fluconazole). In an effort to monitor the emergence of antifungal resistance, standardized culture technique and antimicrobial susceptibility interpretive breakpoints have been developed for Candida sp.¹⁶ As these techniques become widely disseminated, it will become essential to use this information in the management of patients, both in the treatment of established disease as well as in the selection of prophylactic or empiric therapy. Given the excellent bioavailability of certain anti-fungal agents and a large therapeutic index, certain fungae with diminished susceptibility, such as C. glabrata, can be treated with a higher antimicrobial doses (eg 400–800 mg qd for fluconazole).¹⁷ Antimicrobial susceptibility should be performed on all serious Candida infections. Institutional antibiograms demonstrating Candida susceptibility should be integrated into practice management, just as antibiograms have become incorporated into the management of bacterial infections. A recent study of Candida bloodstream isolates from the southeastern US demonstrated C. albicans to be fluconazole and itraconazole resistant 15 and 17% of the time, respectively.¹¹ These data are concerning and raise the specter of further emergence and dissemination of azole-resistant Candida sp.

The second strategy to address fungal disease is empiric therapy in high-risk patients: persistent fever in the setting of neutropenia despite appropriate antibacterial antimicrobials for >96 h. Until recently, the standard approach in this setting has been amphotericin B desoxycholate (dose varies between centers from 0.3 to 1 mg/kg/day) until resolution of neutropenia (ANC 500 cell/ml). This approach is limited by the substantial infusion-related toxicity and nephrotoxicity of this agent. With the recent successful development of other antifungal agents, several alternative approaches in this setting are under investigation. Lipid formulations of amphotericin B appear to have equivalent efficacy compared with desoxycholate, with less toxicity; however, the cost has limited broad use. A recent study with intravenous itraconazole compared to amphotericin B demonstrated equivalent efficacy as empiric antifungal therapy in febrile, neutropenic, cancer patients. Itraconazole was associated with significantly less toxicity.^18,19,20 As empiric therapy in the febrile, neutropenic patient, voriconazole was inferior to amphotericin B; however, the clinical utility of the composite primary end point in this study is questionable, as fewer breakthrough Aspergillus infections occurred in the voriconazole-treated group.²¹ Two recent studies have demonstrated substantial efficacy of voriconazole in the treatment of invasive aspergillosis (IA).^22,23 A large study assessing the safety and efficacy of caspofungin as empiric treatment in the setting of fever and neutropenia is being completed. The data from this study will clarify the role of caspofungin in this setting.

A third strategy for preventing IFI in high-risk patients is pre-emption. This approach minimizes exposing patients without an IFI to the toxicities of antifungal agents. Caillot et al¹⁸ have proposed serial chest CT scans during neutropenia in high-risk patients to detect pulmonary infection prior to clinical symptomatology. A more promising approach to guide pre-emptive therapy is the development of molecular markers. Potential markers under investigation include detection of Aspergillus constituents, such as galactomannan (GM) or glucan synthase, and DNA detection through polymerase chain reaction (PCR), using either pan-fungal primers or species-specific primers.²⁴ In one study of GM as a marker to identify invasive Aspergillus infection, serial GM measurements in 362 consecutive high-risk treatment episodes in 191 patients (BMT and leukemic) were performed, and detection of the presence of GM demonstrated an 85% PPV and 100% NPV for proven and 50% PPV and 98% NPV for probable IA (based on 2 positive GM samples). As the certainty of the diagnosis of IA changes, the testing characteristics also change.^25,26 This test is not yet commercially available in the USA.

PCR diagnosis for IA²⁷ is also a rapidly evolving area in high-risk allogeneic BMT patients. In one study of 84 allogeneic BMT patients, PCR detected Aspergillus 2 days prior to clinical symptoms and 9 days prior to clinical diagnosis. The PCR assay in this small study was 100% sensitivity and 84% specificity.²⁸ It remains to be demonstrated whether earlier diagnosis of IA effects clinical outcome. In another report, PCR on blood to detect Candida and Aspergillus was studied in 42 patients with neutropenia and cancer. In this study assays were performed every 2–3 days, with a positive result, rather than prolonged febrile neutropenia, triggering deployment for antifungal therapy.²⁹ These data are too preliminary to draw firm conclusions; however, the development of this approach is needed to minimize the toxicity of empiric therapy. A major limitation of PCR is background environmental contamination causing false-positive assay results. Two important considerations in the development of specific molecular diagnostic markers are the breadth of pathogens detectable (as this will impact the safety of withholding empiric therapy) and assay performance with various sites of infection (eg CNS disease). When assessing the utility of a molecular marker, one must be clear as to the question: avoidance of unnecessary toxic therapy (negative predictive value), diagnosis of early infection (positive predictive value), vs test of cure or monitoring for relapsed infection. With these new molecular markers, the hope is to develop a pre-emptive strategy analogous to that which has been developed for the management of CMV post-transplantation, where molecular evidence of viral reactivation is seen prior to evidence of infection, allowing earlier intervention with ganciclovir to prevent the development of clinically apparent infection and minimizing unnecessary toxic medication use.³⁰

Optimal therapy for IFI, especially IA, the most common mold infection, needs to be defined. It is well appreciated that an Aspergillus infection in this patient population is often a terminal event. The approach to therapy should include minimization of immunosuppression, augmentation of immunity, surgical excision (if possible), and treatment of permissive viral infections (CMV and human herpesvirus-6A and 6B), which also play a role in augmenting local immunosuppression, in addition to antifungal therapy. The historical standard therapy for IFI, for the last 50 years, is amphotericin B. New classes of antifungal agents are available which need to be incorporated into our management strategies. Flucytosine (a pyrimidine), ketoconazole, fluconazole (FDA approved 1990), itraconazole (FDA approved 1992) are all antifungal agents approved for systemic fungal infection. Lipid formulations of amphotericin B (initial FDA approval in 1995) were developed to minimize the toxicities associated with amphotericin B desoxycholate and the first echinocandin, caspofungin, was recently FDA approved in 2001. Of these agents, flucytosine is rarely used because of its limited efficacy and bone marrow toxicity. Itraconazole had limited utility in the febrile, neutropenic setting because of poor oral bioavailability, until an intravenous formulation became available last year.³¹ Fluconazole is not active against Aspergillus. The role of caspofungin is undefined in this setting because of limited data at this time (Maertens J et al., 40th ICAAC, 2000; Abstract 1103).

These three drug classes (polyene, azole, and echinocandins) have different mechanisms of action. Polyenes bind ergosterol and cause perforation in the fungal cell wall, which in turn leads to cell death. Azoles inhibit synthesis of ergosterol.³² Given these two classes act on the same pathway, there is theoretical concern for antagonism when these agents are used in combination. Antagonism was demonstrated in an animal model, when ketoconazole, followed by amphotericin B, was administered in a rodent model of Aspergillus endocarditis.¹ However, recent clinical data have not demonstrated antagonism in the treatment of candidemia (Rex JH et al. 41st ICAAC 2001; Abstract J-681a). Other new azole antifungal agents include voriconazole (recently FDA approved),^22,23 and posaconazole and ravuconazole, which are in advanced clinical development. It is important to note that, voriconazole has typically been used as primary or salvage therapy for IA, thus little clinical data are available as to the potential for clinically relevant anatagonism, if azole administration is followed by polyene therapy. Echinocandins act by inhibiting beta-1,3 glucan synthase, an important component of the fungal cell wall, thus obviating the above-mentioned concern. An important limitation of the echinocandin is the requirement for intravenous administration. Anidulafungin (LY 303366) and micafungin (FK 463) are other echinocandins in advanced clinical development.

Given the generally poor outcome of IA in this patient population, approximately 10–20% success, the potential of combination therapy requires examination. What to use, at what dose, and in which order are unknown at this time. As antifungal agents will be administered with other supportive medications, such as immunosuppressants, growth factors, and corticosteroids, drug–drug interactions are a major concern. For example, voriconazole inhibits the P-450 enzyme system, thus leading to increased levels of cyclosporine (two-fold), tacrolimus (three-fold), and rapamycin (10-fold).³³ Given the substantial decrease in rapamycin clearance, voriconazole should be used in this setting with the utmost care.

Duration of therapy in a patient diagnosed with an IFI depends on several key factors: degree of immunosuppression, presence and severity of GvHD, control of the underlying malignancy, and burden and site(s) of infection. Given the inherent variability of these factors, it is difficult to estimate how long to treat a fungal infection; however, the typical course of therapy is months to years. The availability of reliable molecular markers, such as PCR for Aspergillus or GM, would allow improved follow-up and management of these patients, as a response to therapy with a decrease in GM has been suggested.²⁶ In addition, relapsing disease may be associated with an increase in GM.²⁶ Some of these drugs, such as voriconazole, can be taken orally on an outpatient basis, thus allowing prolonged therapy.

Susceptibility testing of yeasts is necessary to determine the emergence of resistant organisms and to guide therapeutic decisions. As the risk period for IFI shifts with GvHD, our approach to prophylaxis and pre-emption needs to be re-evaluated. Pre-emptive monitoring with novel biomarkers may trigger pre-emptive or empiric therapy (eg in the setting of fever and neutropenia), thus minimizing medication-induced toxicity. Which agent to use for prevention and treatment of IFI will evolve as data with the new therapies become available.