Recent developments in the management of invasive fungal infections in patients with oncohematological diseases

Abstract

Patients with hematological cancer have a high risk of invasive fungal diseases (IFDs). These infections are mostly life threatening and an early diagnosis and initiation of appropriate antifungal therapy are essential for the clinical outcome. Most commonly, Aspergillus and Candida species are involved. However, other non-Aspergillus molds are increasingly be identified in cases of documented IFDs. Important risk factors are long lasting granulocytopenia with neutrophil counts below 500/μl for more than 10 days or graft-versus-host disease resulting from allogeneic stem-cell transplantation. For definite diagnosis of IFD, various diagnostic tools have to be applied, including conventional mycological culture and nonconventional microbiological tests such as antibody/antigen and molecular tests, as well as histopathology and radiology. In the last few years, various laboratory methods, like the Aspergillus GM immunoassay (Aspergillus GM EIA), 1,3-ß-D-glucan (BG) assay or polymerase chain reaction (PCR) techniques have been developed for better diagnosis. Since no single indirect test, including radiological methods, provides the definite diagnosis of an invasive fungal infection, the combination of different diagnostic procedures, which include microbiological cultures, histological, serological and molecular methods like PCR together with the pattern of clinical presentation, may currently be the best strategy for the prompt diagnosis, initiation and monitoring of IFDs. Early start of antifungal therapy is mandatory, but clinical diagnostics often do not provide clear evidence of IFD. Integrated care pathways have been proposed for management and therapy of IFDs with either the diagnostic driven strategy using the preemptive antifungal therapy as opposed to the clinical or empirical driven strategy using the ‘traditional’ empirical antifungal therapy. Antifungal agents preferentially used for systemic therapy of invasive fungal infections are amphotericin B preparations, fluconazole, voriconazole, posaconazole, caspofungin, anidulafungin, micafungin, and most recently isavuconazole. Clinical decision making must consider licensing status, local experience and availability, pharmacological and economic aspects.

Introduction

Immunocompromised patients, in particular those with cancer or hematological malignancies and allogeneic bone marrow/hematopoietic stem-cell transplant recipients (BMT/HSCT) carry a high risk of invasive fungal diseases (IFD), which are frequently life threatening. An early diagnosis and prompt initiation of appropriate antifungal therapy are imperative and essential for a favorable clinical outcome. However, an early diagnosis is usually difficult to establish, due to the limited value of microscopy, mycological cultures and conventional radiology. Clinical features of an IFD like fever, shortness of breath or thoracic pain are nonspecific. Furthermore, fever may be absent in patients with corticosteroid therapy and the so-called ‘typical’ radiological signs (e.g. ‘halo sign’) are rather late features of an advanced infection or may fail to appear in patients with granulocytopenia [Caillot et al. 2001]. In general, the interval between subtle fungal tissue invasion and appearance of clinical signs and symptoms determines the time point of initiation of antifungal therapy and may have prognostic implications. Initiation of systemic antifungal therapy prior to establishing the diagnosis of a fungal infection may contribute to a better therapeutic outcome in high-risk patients.

Around 30% of patients not on mold-active prophylaxis who develop fever of unknown origin (FUO) during granulocytopenia have IFD. However, administration of empirical therapy to all persistently febrile patients as defined above may lead to overtreatment with expensive and potentially toxic drugs in a relevant proportion of patients [de Pauw, 2005]. Moving from empirical to preemptive therapy may reduce the number of patients with antifungal overtreatment [de Pauw and Rubin, 2006]. Preemptive therapy is considered in patients with persistent FUO refractory to broad-spectrum antibacterial therapy during granulocytopenia and clinical or laboratory/radiological signs supporting the diagnosis of an IFD [e.g. detection of Aspergillus galactomannan (GM) antigen or ‘halo’ sign on chest computed tomography (CT) scan] [Greene et al. 2007; Maertens et al. 2005]. This approach might allow treating early ‘probable’ infections [Morrissey et al. 2013, 2014].

Furthermore, this would avoid overtreatment in a subset of high-risk patients, avoiding treatment-related toxicities, preventing growth of resistant species and reducing resource expenses [de Pauw, 2005]. Recent developments in diagnostic and therapeutic strategies of IFD in patients with hematological malignancies are discussed in this article.

Patients at risk of invasive fungal diseases

Invasive candidosis, particularly blood stream infections leading to candidemia, represents the most frequent systemic fungal infection in patients in the surgical intensive care unit (ICU) undergoing complex abdominal surgery (e.g. for bowel perforation), general ICU patients with multiorgan failure and high severity of illness score (e.g. APACHE II/III score) [Pappas et al. 2016; Ruhnke et al. 2011]. Various other patients and conditions are associated with an increased risk of IFD, such as patients receiving total parenteral nutrition, having central-venous catheters, patients with granulocytopenia and malignancies, burn patients, low-weight premature infants, patients receiving long-term treatment with more than 20 mg of prednisone per day or other immunosuppressive drugs (e.g. anti-tumor necrosis factor α inhibitors) and prolonged treatment with broad-spectrum antibiotics [Pappas et al. 2009; Ruhnke et al. 2011].

In contrast, invasive aspergillosis (IA), mostly invasive pulmonary or disseminated aspergillosis, occurs primarily in patients with acute leukemia and patients with prolonged granulocytopenia due to hematological malignancies, as well as in patients undergoing allogeneic bone marrow or peripheral blood stem-cell transplantation with graft-versus-host disease [Cordonnier et al. 2006; Pagano et al. 2006; Patterson et al. 2000]. In a prospective surveillance study for IFD in HSCT recipients in 2001–2006 from the USA (Transnet), IA (43%), invasive candidosis (28%) and zygomycosis (8%) were the most common IFDs [Kontoyiannis et al. 2010]. Beyond this, a broad spectrum of other immunocompromised patients, such as solid organ transplant recipients, patients with inborn immunodeficiencies, patients with autoimmune diseases receiving immunosuppressive medication, chronic pulmonary diseases, and selected patients in the ICU are at risk of acquiring severe fungal mold infections [Walsh et al. 2008]. Other mold infections such as zygomycosis (or mucormycosis) may occur in distinct patient groups other than in patients with hematological malignancies (e.g. diabetes mellitus or trauma) as well [Lanternier et al. 2012; Petrikkos et al. 2012]. Recently, a risk prediction score for invasive mold disease (IMD) in high-risk patients was proposed (with four risk factors such as granulocytopenia, lymphocytopenia or lymphocyte dysfunction in allogeneic HSCT recipients, malignancy status and prior IMD retained in the final multivariate model) to distinguish patients at low risk of developing IMD [Stanzani et al. 2013]. Clinical signs and symptoms associated with invasive fungal infections in patients with hematological malignancies are listed in Table 1.

Table 1.

Clinical signs and symptoms associated with invasive fungal infections (modified according to Denning et al. [1997]).

Organ	Symptoms	Diagnosis
Skin	Papules, nodules, pustules, vesicles, purpura, ulcer, cellulitis, superficial granulomas or plaques, abscesses, necrotic skin lesions	Acute disseminated candidosis, disseminated aspergillosis or fusariosis, cryptococcosis (and other mould infections)
Sinonasal	Facial pain, epistaxis, nasal congestion, crusting or ulcers, necrotic eschar	Invasive aspergillosis, invasive zygomycosis (and other mold infections)
Oral cavity	Erythematous or whitish spots/lesions necrotic eschar/lesions or ulcers	Oral candidosis, invasive aspergillosis, invasive zygomycosis
Chest	Pleuritic chest pain, dyspnea, non-productive cough, hemoptysis, pleural rub cough, chest pain	Invasive aspergillosis and other mold infections, DD Pneumocystis jirovecii, pneumonia
Eye	Chorioretinitis/endophthalmitis during fundoscopy ‘cotton wool’-like lesions	Ocular or acute disseminated candidosis, rarely mold infection
CNS	Headache, meningism, focal neurological deficit, seizures, altered mental status	Meningoencephalitis due to cryptococcosis, DD Candida meningitis, abscesses due to cerebral mold infection

DD, differential diagnosis.

Pathogens

The most common cause of nosocomial fungal diseases for all patient populations that are immunocompromised are yeast pathogens, in particular Candida albicans, followed by non-C. albicans spp. (e.g. C. glabrata, C. krusei, C. parapsilosis, C. tropicalis) [Pagano et al. 2006; Ruhnke, 2006; Viscoli et al. 1999]. Second, mold pathogens like Aspergillus fumigatus (>80%) and other Aspergillus spp. (e.g. A. terreus, A. niger, A. nidulans in <5% for each pathogen) are responsible for invasive fungal infections. IA, in particular invasive pulmonary aspergillosis (IPA) as well as invasive candidosis, in particular candidemia, are the most frequent clinical manifestations of fungal pathogens in immunocompromised patients [Pagano et al. 2006]. In addition to the more common fungal infections, caused by Candida and Aspergillus spp., there are growing numbers of fungal infections caused by zygomycetes (Mucorales spp. and others), Fusarium spp., Trichosporon spp., Cryptoccocus neoformans, rare fungi such as black molds (e.g. Alternaria, Curvularia, Bipolaris spp.) and others reported from some hematological centers [Chamilos et al. 2005; Kontoyiannis et al. 2004a, 2004b, 2005; Krcmery et al. 1999; Pagano et al. 2007]. Infections due to Pneumocystis jirovecii have been occasionally described in hematological patients but will not be further discussed in this review [Li et al. 2014]. Most data on diagnosis and treatment of cryptococcal meningitis have been obtained from patients with acquired immune deficiency syndrome. Reports from patients who are human immunodeficiency virus negative with hematological disorders are limited [Pagano et al. 2004]. According to an epidemiological study from Italy in a cohort of 11,802 patients with hematologic malignancies, there were 538 proven or probable IFDs (4.6%) [Pagano et al. 2006]. Of these, the majority of infections (346/538) were caused by molds (64%), in most cases Aspergillus spp. (310/346). The majority of yeast infections were cases of candidemia (175/192). The highest IFD-attributable mortality rates were associated with zygomycosis (64%) followed by fusariosis (53%), aspergillosis (42%) and candidemia (33%). Recently, the first data of a European period-prevalence study to estimate the rate of invasive pulmonary mold disease (PIMDA study) were presented [Donnelly et al. 2015]. Information is available online and full-paper publication is pending.

Diagnosis

The classification of IFDs is based on the level of diagnostic certainty and includes proven, probable and possible IFD according to the international consensus criteria of the ‘Invasive Fungal Infections Cooperative Group’ of the ‘European Organization for Research and Treatment of Cancer’ (EORTC) and the ‘Mycoses Study Group’ (MSG) [de Pauw et al. 2008]. These diagnostic criteria were developed particularly for clinical studies to standardize the target groups of patients. Because of their strict diagnostic requirements, these definitions might be helpful in the clinical routine, but they are not intended to guide therapeutic decisions in individual patients. According to these criteria, a definite diagnosis of IFD can be only provided by cultures obtained under aseptic conditions from otherwise sterile sites (e.g. blood cultures) or histological specimens demonstrating tissue invasion.

Culture-based mycological diagnosis

Blood cultures are the method of choice for the diagnosis of candidemia and other yeast infections. Standard blood culture media detect most Candida spp. and other yeasts. The addition of special fungal media may further enhance the speed and recovery of yeasts from blood (‘Mycosis-IC/F-Medium’ or BacT/ALERT 3D) [Fricker-Hidalgo et al. 2004; Horvath et al. 2004, 2007]. However, a separate blood culture bottle has to be used for this procedure. At least two pairs of blood culture bottles (10 ml each) should be obtained for aerobic and anaerobic culture when candidemia is suspected before the initiation of antifungal therapy [Ruhnke et al. 2012]. Using this approach, up to 75–90% of candidemia episodes can be detected. To increase the yield of blood cultures above 90%, up to four blood culture pairs have to be taken from various body sites [Lee et al. 2007]. The European Society for Clinical Microbiology and Infectious Diseases guideline for the diagnosis and management of Candida diseases recommends performing daily blood cultures until culture results turn negative to monitor response to antifungal treatment and to determine the appropriate duration of antifungal therapy, which should be continued for a minimum of 14 days after the end of candidemia [Cornely et al. 2012; Ullmann et al. 2012].

In the majority of patients with pulmonary fungal infections diagnostic confirmation is based on respiratory samples [e.g. sputum, tracheal secretions, bronchoalveolar lavage (BAL)]. Increasing the number of sputum samples could improve the diagnostic sensitivity, with three samples providing optimum yield in IA, as suggested by recommendations from the European Conference in Infections in Leukemia meetings [Arendrup et al. 2012]. BAL fluid provides a more representative sample from the lower respiratory tract and allows CT scan abnormalities to be directly sampled, but the overall sensitivity of culture and microscopy for the diagnosis of IA in the hematology population is probably not higher than 50% [Arendrup et al. 2012; Maschmeyer et al. 2015; Ruhnke et al. 2012].

In vitro susceptibility testing (minimal inhibitory concentration is indicated for all fungal isolates from blood and other sterile specimens) [Arendrup et al. 2012]. However, breakpoints for in vitro susceptibility are not yet standardized for all antifungals. Susceptibility testing of Candida isolates and A. fumigatus isolates (and other Aspergillus spp.) has gained interest because of the emergence of azole-resistant A. fumigatus isolates from immunocompromised patients [Snelders et al. 2008]. Multiple mechanisms of resistance have been identified, with different degrees of azole cross resistance, including mutations in the cytochrome P450 51A gene at G54, L98+TR, G138, M220, G448 [Verweij et al. 2009]. Positive cultures with growth of Aspergillus spp. in patients with hematological malignancies is uncommon and molecular diagnostic tests have been developed to detect these mutations in body fluids (e.g. BAL) [Spiess et al. 2014]. However, the clinical impact of molecular resistance analyses of A. fumigatus in patients with hematological malignancies needs to be established in clinical studies.

Data on the epidemiology of azole-resistant A. fumigates isolates in patients with hematological cancer are scarce and may differ from country to country [Buchheidt et al. 2014; Kidd et al. 2015; Snelders et al. 2008]. According to a recent prospective multicenter international surveillance study analyzing 3788 Aspergillus isolates in 22 centers from 19 countries, the prevalence of azole-resistant A. fumigatus was found to be 3.2% [van der Linden et al. 2015]. This issue may become clinically important in the future because in vitro resistance to mold-active azole antifungals (e.g. itraconazole, voriconzole or posaconazole) has been associated with clinical treatment failure and mortality under therapy with voriconazole [Rath et al. 2012; Thors et al. 2011; Verweij et al. 2007]. In a recent study from the Netherlands, the case-fatality rate in patients with azole-resistant IA was 88.0% [van der Linden et al. 2011].

Nonculture-based mycological diagnosis

Nonculture-based markers of IFD like circulating GM or BG play an important diagnostic role. Fungus-specific antibodies and antigens may be detected by noncultural methods even prior to the clinical or radiological signs of IFD [Herbrecht et al. 2002]. The detection of GM or BG in body fluids is included in the list of microbiological criteria of the EORTC/MSG definitions for diagnosing IA [de Pauw et al. 2008]. A high negative predictive value (NPV) of over 95% with a threshold of 0.5 (serum galactomannan index) have been reported for the GM sandwich-enzyme immunoassay (Aspergillus GM ELISA), demonstrating that this technique may be used to make the diagnosis of IA unlikely [Herbrecht et al. 2002; Maertens et al. 2007; Marr et al. 2004a; Pfeiffer et al. 2006]. Furthermore, detection of GM may be used to monitor treatment responses, since GM concentrations may correspond with the fungal burden [Chai et al. 2012]. Higher serum GM was associated with higher overall mortality in two cohorts of allogeneic HSCT recipients [Fisher et al. 2013; Mikulska et al. 2013]. In contrast, sensitivity of this assay may be decreased while the patient is already on treatment with a mold-active antifungal drug and may lead to false-negative results [Marr et al. 2004a, 2005; Marr, 2008]. In contrast, the GM test may yield false-positive results for various reasons, including administration of GM-containing batches of β-lactam antibiotics [Mennink-Kersten et al. 2004; Viscoli et al. 2004].

BG is a cell wall component, which may circulate in the blood of patients with various types of invasive fungal infections and is therefore a useful ‘broad-spectrum’ fungal marker. Investigations using different BG assays have reported a sensitivity of 55–100%, specificity of 87–93%, positive predictive values of 40–84%, and NPVs ranging from 75% to 100% [Odabasi et al. 2004; Ostrosky-Zeichner et al. 2005; Pickering et al. 2005].

However, BG testing does not allow differentiation of yeast from mold infections and its potential to produce false-positive results (e.g. due to glucan-containing dialysis membranes, immunoglobulins or concomitant Gram-positive bacteremia) limits its clinical utility [Marchetti et al. 2011; Pickering et al. 2005]. Serological (antibody) test are currently not recommended for the diagnosis of invasive candidosis or candidemia in hematological patients due to lack of clinical validation [Marchetti et al. 2011; Ruhnke et al. 2011]. Only a combined testing of Candida antigens and antibodies may contribute to the diagnosis of a Candida infection, in particular for diagnosis of hepatosplenic candidosis according to the current European Conference on Infections in Leukemia (ECIL) recommendations [Marchetti et al. 2011; Mikulska et al. 2010; Sendid et al. 2002].

Several methods have been developed for detecting fungal DNA in clinical specimens by nucleic acid hybridization and amplification. The specificity varies from 65% to 75%, depending on the number of tests needed to establish the diagnosis with a 100% sensitivity and results from BAL appear to be superior compared with blood specimens testing [Buchheidt et al. 2004]. A ‘Panfungus’, PCR allows detection of the DNA specific for most pathogenic fungi, whereas real-time PCR assays are more species specific [Donnelly, 2006]. A multicenter study evaluating the utility of fungal PCR (Candida and Aspergillus spp.) has been successfully performed in the UK and Ireland [White et al. 2006]. However, since no common methodological guidelines have been established until recently, PCR testing has not yet been integrated into the EORTC/MSG definitions [de Pauw et al. 2008]. A lack of standardizing of in-house PCR systems has prompted a European initiative for standardization of A. fumigatus PCR [White et al. 2010]. This international working group could effectively harmonize the various PCR protocols in large multicenter laboratory studies by establishing a worldwide standard which may allow Aspergillus PCR to be included in upcoming EORTC/MSG definitions [Springer et al. 2013; White et al. 2015].

In addition, new diagnostic tests [e.g. lateral flow device (LFD); volatile metabolite profile of A. fumigatus in breath] have been developed and initial data are very promising [Hoenigl et al. 2014; Koo et al. 2014; White et al. 2013]. However, it is unclear whether these tests will come to be commercially available in the near future. Currently, the best diagnostic accuracy can be achieved using a combination of noncultural test (e.g. Aspergillus GM + PCR ± LFD) for diagnosing IPA in BAL from patients with hematological malignancies [Hoenigl et al. 2014; White et al. 2015].

Radiology

In contrast, radiology imaging plays an important role in the diagnosis of IFD, in particular CT scans [Heussel et al. 1997, 1999]. Conventional chest X rays are of limited value for early diagnosis of IFD in high-risk patients because lack of sensitivity during granulocytopenia [Heussel et al. 1999; Oude Nijhuis et al. 2003]. The typical radiological ‘halo’ or ‘air-crescent’ sign, which is typically seen in IA and is best visualized by chest CT, either with multislice thoracic CT or high-resolution CT, is not entirely specific for an IFD. However, it is clearly defined and allows the diagnosis of probable or possible IFD according to the EORTC/MSG criteria and consequently presents a basis for preemptive antifungal therapy [Caillot et al. 2001; de Pauw et al. 2008; Greene et al. 2007]. Cavitations and the air-crescent sign typically occur during or after recovery from granulocytopenia [Brodoefel et al. 2006; Kim et al. 1999]. The CT findings of consolidation, as well as the ‘air-crescent’ and ‘halo’ sign, were regarded as major diagnostic criteria, while any new infiltrate was classified as a minor sign for diagnosing fungal infection in the previous EORTC/MSG consensus criteria [Ascioglu et al. 2002]. In the updated definitions, there is no discrimination between major and minor clinical criteria [de Pauw et al. 2008].

The dense, well circumscribed lesion on CT may or may not carry a ‘halo’ sign and may be considered as a diagnostic sign of an invasive pulmonary fungal infection. However, in patients with granulocytopenia the detection of the ‘halo’ sign on chest CT is routinely used by most hematology centers as a trigger for initiation of preemptive antifungal therapy. In recent years, a ‘reversed halo sign’ (RHS) using CT scans has been described. The RHS is defined by a focal round area of ground-glass attenuation surrounded by a crescent or ring of consolidation [Marchiori et al. 2012b]. The RHS has been described more commonly with pulmonary zygomycosis (or mucormycosis), but can also be present in IPA and other pulmonary infections [Marchiori et al. 2012a, 2012b].

Most recently, the diagnostic performance of CT pulmonary angiography (CTPA) versus other CT imaging findings was analyzed in 100 patients with hematological malignancies and possible IMD [Stanzani et al. 2015]. In total, 46/100 patients who underwent CTPA were upgraded to probable or proven mold disease. The authors concluded that vessel occlusion detected by CTPA is a more sensitive and possibly more specific radiographic sign than other forms of CT. However, the authors used high-resolution CT as a comparator, which may not be as sensitive as multislice chest CT. It remains to be established whether this method is applicable in clinical routine.

¹⁸F-fluorodeoxyglucose positron emission tomography (¹⁸F-FDG PET±CT) may be helpful, particularly to rule out undetected infection [Vos et al. 2012]. PET CT was explored in a small study of patients with febrile neutropenia as an adjunct to conventional evaluation and management but not in a larger cohort of hematological patients with IFD [Guy et al. 2012]. The role of ¹⁸F-FDG PET±CT scans has been found helpful in individual cases with hepatosplenic candidosis, but cannot distinguish between malignancy and infection [Teyton et al. 2009].

Antifungal treatment

Diagnostic-driven or preemptive therapy

The preemptive approach is based on clinical or laboratory findings, suggesting an IFD without pathogen identification from sterile site material (such as blood culture) or tissue specimens. This strategy avoids the unnecessary usage of antifungal drugs and avoids drug-related toxicities in patients with nonfungal causes of the fever, but it allows early therapy in patients who are lacking early symptoms of a fungal infection. As a marker for measuring ‘fungal load’ in blood specimens, antigen tests have been developed such as the Aspergillus GM but also molecular tests (PCR) for detection of Aspergillus spp.. It has been shown in a prospective clinical study that preemptive antifungal therapy guided by sequential monitoring of Aspergillus GM (e.g. daily) together with high-resolution CT reduced the exposure to antifungal drugs [Maertens et al. 2005]. In a multicenter, open-label, randomized trial from France, empirical antifungal therapy was compared with a preemptive strategy [Cordonnier et al. 2009]. Preemptive therapy was defined as treatment of patients who had clinical, imaging or GM antigen assay evidence suggesting IFD. Probable or proven invasive fungal infections were more frequent among patients with preemptive treatment compared with patients who received empirical treatment. However, overall survival as the primary endpoint was similar in both patient groups. Recently, data from an open-label, parallel-group, randomized controlled trial of 240 patients undergoing allogeneic stem-cell transplantation or chemotherapy for acute leukemia, with no history of IFD, were published from Australia [Morrissey et al. 2013]. Patients were randomized to a standard diagnostic strategy (based on culture and histology) or a biomarker-based diagnostic strategy (Aspergillus GM and PCR) to direct treatment with antifungal drugs. Thirty-two percent of patients in the standard diagnosis group and 15% in the biomarker diagnosis group received empirical antifungal treatment (p = 0.002). The authors concluded from their data that Aspergillus GM and PCR to direct treatment reduced use of empirical antifungal treatment. As a result, from a European consensus meeting integrated pathways were proposed to guide clinicians for diagnostic-driven antifungal therapy of IMD [Agrawal et al. 2011].

Targeted therapy

Invasive candidosis

Invasive candididosis in patients with cancer is primarily caused by C. albicans, although an increase of candidemia by non-albicans Candida spp. has been reported from some centers. Effective antifungal treatment should be started in patients with granulocytopenia as soon as possible in order to minimize fungal-related fatality [Mousset et al. 2013; Ullmann et al. 2012]. Therefore, exact species identification should not be awaited, but treatment should be initiated as soon as the finding of yeasts isolated from blood culture is reported. The gastrointestinal tract should be considered as another main source of invasive candidosis in patients with granulocytopenia. In the case of a proven fluconazole-sensitive invasive Candida infection, stepping down from a broader-spectrum systemic antifungal to oral fluconazole has been shown to be feasible [Vazquez et al. 2014]. The recommended duration of treatment is at least until 14 days after the first negative blood culture and resolution of signs and symptoms. Central venous catheter-related invasive candidosis should prompt catheter removal [Mousset et al. 2013; Ullmann et al. 2012]. For treatment in these cases, echinocandins or liposomal amphotericin B (L-AmB) appear more effective against biofilms than azoles or conventional amphotericin. In general, echinocandins or L-AmB are the preferred antifungal agents for treating invasive Candida infections in patients with granulocytopenia [Andes et al. 2012; Maertens et al. 2011; Mousset et al. 2013; Ullmann et al. 2012]. In patients with fever persisting despite neutrophil recovery, hepatosplenic candidosis should be considered. AmB formulations, fluconazole or caspofungin are accepted treatment options, the latter being preferred in cases where the Candida spp. is not identified. Antifungal therapy should be given until calcification or complete remission of hepatosplenic lesions. However, in single patients refractory fever may be caused by immune reconstitution susceptible to corticosteroid treatment [Legrand et al. 2008]. In patients with Candida infections of the central nervous system, usually caused by hematogenic spread, treatment options are less well defined. From pharmacological considerations, L-AmB, with or without 5-flucytosine or fluconazole, may be preferred, and due to its improved penetration into the cerebrospinal fluid, voriconazole appears to be a potent alternative [Pappas et al. 2016; Ruhnke et al. 2011]. Treatment should be continued for an additional 4 weeks following the resolution of manifestations. Brain abscesses may require drainage or, if feasible, surgical resection. For Candida infection of the urinary tract, fluconazole is preferred in cases of susceptible Candida isolates, and urine catheters should be removed whenever clinically feasible. Table 2 shows current treatment recommendations based on results from clinical studies.

Table 2.

Antifungal therapy of selected invasive fungal infections in hemato-oncological patients (modified according to Mousset et al. [2013]; strength of recommendation (SR) and grades indicating the quality of evidence (QE) according to the criteria of the IDSA [Kish, 2001]).

Invasive aspergillosis	SR/QE#	Invasive candidosis	SR/QE#
First-line treatment of IPA		Candidemia
Voriconazole	AI	Granulocytopenic patient
Isavuconazole	AI	Echinocandins	BI
Liposomal AmB	AII	Liposomal AmB	BI
Caspofungin	CII	Nongranulocytopenic patient
Micafungin	CII	Echinocandins	AI
ABLC	CIII	Liposomal AmB	AI
Anidulafungin + voriconazole	CIII	Voriconazole	AI
D-AmB	EI	Fluconazole	AI
Second-line treatment of IPA		Hepatosplenic candidosis
Voriconazole	BII	Granulocytopenic patient
Caspofungin	BII	Liposomal AmB/ABLC	BIII
Posaconazole	BII	Echinocandins	BIII
Liposomal AmB	BII	Voriconazole	CIII
ABLC	BII	Nongranulocytopenic patient
Micafungin	CIII	Fluconazole	BIII
Voriconazole + caspofungin	CIII
		Steroids (as adjunctive therapy)	CIII
CNS infection/sinusitis
Voriconazole	AII
Liposomal AmB (⩾5 mg/kg)	BIII
Surgical intervention (+AF)	AII
Invasive mucormycosis	SR/QE#	Invasive mucormycosis	SR/QE#
First-line treatment		Second-line treatment
Liposomal AmB (⩾5 mg/kg)	AII	Isavuconazole	AII
Isavuconazole	AII	Posaconazole	AII
ABLC	BII	Liposomal AmB (⩾5 mg/kg)	BII
Posaconazole	BII	ABLC	BII
Lipid-based AmB + caspofungin	CIII	Surgical intervention (+AF)	AII

ABLC, amphotericin B lipid complex; AF, antifungal therapy; AmB, amphotericin B; CNS, central nervous system; D-AmB, Amphotericin B deoxycholate; IDSA, Infectious Disease Society of America; IPA, invasive pulmonary aspergillosis.

Invasive aspergillosis

In patients with granulocytopenia and IA, the lungs (IPA) are involved in up to 90% of cases, whereas sinusitis, disseminated infections and cerebral involvement are much less common. To improve the fatality rate, early treatment at first signs of infection, typically based on pulmonary CT findings in patients with febrile granulocytopenia, is mandatory [Maertens et al. 2011; Mousset et al. 2013]. For primary treatment of IPA, voriconazole and L-AmB are preferred [Maertens et al. 2011; Mousset et al. 2013]. Because of its sufficient penetration into the central nervous system, voriconazole is recommended for primary treatment of cerebral aspergillosis [Schwartz et al. 2005]. Due to its unfavorable toxicity profile and unsatisfactory efficacy, conventional AmB should be avoided in IA [Mousset et al. 2013; Schwartz et al. 2007]. If feasible, surgical resection of singular lesions is recommended. Recently, data from a large phase III, double-blind, global multicenter, comparative group study (SECURE trial) in 527 adult patients comparing isavuconazole and voriconazole were published [Maertens et al. 2016]. Isavuconazole was found to be equally effective as voriconazole in invasive mold infections (mostly IA) but associated with a better toxicity profile. Isavuconazole-treated patients had a lower frequency of hepatobiliary disorders [23 (9%) versus 42 (16%)], eye disorders [39 (15%) versus 69 (27%)] and skin or subcutaneous tissue disorders compared with voriconazole [Maertens et al. 2016].

The potential benefit of combination therapy in IA is a matter of controversy [Chamilos and Kontoyiannis, 2006]. Combination therapy is recommended in the IDSA guidelines for second-line (salvage) therapy [Walsh et al. 2008] and combination therapy is preferably done using two or more agents with different modes of action (mostly azole or polyene ± echinocandin). The combination of voriconazole and caspofungin as second-line therapy in HSCT recipients showed a better outcome compared with a historical control group who were treated with voriconazole alone [Marr et al. 2004b]. The combination of caspofungin with L-AmB has been studied in a smaller trial and this combination resulted in a 42% response rate in patients who frequently failed to respond to monotherapy [Kontoyiannis et al. 2003]. In a prospective pilot study, comparing the combination of L-AmB (3 mg/kg per day) with caspofungin versus high-dose L-AmB (10 mg/kg per day) alone, the combination therapy resulted in a better response [Cornely et al. 2007; de Pauw et al. 2008]. However, in this study the response rate in patients with L-AmB monotherapy was remarkably low at only 27%. Micafungin has shown a good activity against Aspergillus in vitro and in animal models in combination with voriconazole, but it has been used clinically only in uncontrolled studies and in combination with various antifungal compounds [Denning et al. 2006; Kontoyiannis et al. 2008; Lewis and Kontoyiannis, 2005]. Furthermore, the efficacy of micafungin in IA as monotherapy has not yet been conclusively demonstrated [Denning et al. 2006]. In cerebral aspergillosis, an animal model suggests an improved efficacy of the combination of voriconazole plus L-AmB compared with either monotherapy [Clemons et al. 2005]. Recently, data from the first (and only) randomized, double-blind, placebo-controlled multicenter trial on the safety and efficacy of voriconazole and anidulafungin compared with voriconazole monotherapy for treatment of IA were published [Marr et al. 2015]. In this study, 454 patients with hematologic malignancies or HSCT and suspected or documented IA were randomly assigned to treatment. The primary outcome was 6-week mortality. Mortality rates at 6 weeks were 19.3% for combination therapy and 27.5% for monotherapy. However, this difference was not statistically different. Interestingly, mortality was lower with combination therapy in a subgroup of patients who received the diagnosis of an IA based on radiographic findings and GM positivity. In patients with Aspergillus sinusitis, surgical intervention is indicated in combination with systemic antifungal therapy [Mousset et al. 2013]. For treatment of breakthrough aspergillosis emerging in patients undergoing posaconazole or voriconazole prophylaxis, a switch to another class of antifungal agents, preferably L-AmB, is justified.

In general, antifungal therapy should be continued until manifestations of IA are completely resolved or reduced to residual scarring. First clinical response assessment should be done after a minimum of 14 days with full-dose treatment. Apart from clearly evident failure due to resistance of the pathogen (e.g. A. terreus to AmB preparations), lack of adequate drug levels at the site of infection, intolerance or severe organ toxicity, nonresponse of IA to an established antifungal therapy within the first 14 days after treatment initiation should be stated with caution. A temporary increase in the volume of pulmonary lesions during the first week of treatment or neutrophil recovery is commonly observed and should not be misinterpreted as antifungal treatment failure. Treatment recommendations are specified in Table 2.

Invasive zygomycosis (mucormycosis)

Invasive fungal infections caused by zygomycetes may be difficult to distinguish from IA or other mold diseases based upon clinical, imaging or laboratory findings. They are markedly less frequent than the latter, and preferably affect the lower respiratory tract, paranasal sinuses and central nervous system. Small case series have been reported from patients with hematological malignancies [Pagano et al. 1997]. After the introduction of voriconazole, an antifungal agent not active against zygomycetes, breakthrough infections due to zygomycetes have been reported mostly after prophylaxis against aspergillosis in small case series but not in larger clinical studies [Cordonnier et al. 2010; Kontoyiannis et al. 2005; Pagano et al. 2005]. Options for first-line chemotherapy of zygomycosis include L-AmB or AmB lipid complex as first choice, given in daily dosages of at least 5 mg/kg [Cornely et al. 2014; Skiada et al. 2013]. Posaconazole and combination therapy of L-AmB or AmB lipid complex with caspofungin are the potential options for second line-treatment. The combination polyene–caspofungin was shown to be effective in the treatment of rhino-orbital-cerebral mucormycosis in nonhematological patients [Reed et al. 2008]. However, randomized studies on this subject have not been conducted. A combination of L-AmB and the iron chelator deferasirox for the treatment of mucormycosis has shown inferior clinical results for the combination compared with the antifungal agent alone [Spellberg et al. 2012a, 2012b]. Most recently, isavuconazole was licensed for treatment of invasive mold infections. Isavuconazole has activity against a number of clinically important yeasts and molds, including Candida spp., Aspergillus spp., Cryptococcus neoformans and Trichosporon spp. and variable activity against the Mucorales [Pettit and Carver, 2015]. A single-arm open-label trial (‘VITAL study’) that studied primary as well as salvage therapy of invasive mucormycosis showed efficacy with isavuconazole that was similar to that reported for AmB and posaconazole [Marty et al. 2016]. In this study, 37 patients with mucormycosis received isavuconazole for a median of 84 days (range 2–882). By day 42, 4 patients (11%) had a partial response, 16 (43%) had stable IFD, one (3%) had IFD progression and 13 (35%) had died. Treatment recommendations are specified in Table 2.

Surgical resection should be considered, whenever feasible for the patient, for rhinocerebral and skin and soft tissue disease. Reversal of underlying risk factors (diabetes control, neutrophil recovery, discontinuation/taper of glucocorticosteroids, reduction of immunosuppressants, and discontinuation of iron chelating agents) is important in the treatment of mucormycosis. The duration of antifungal therapy is not defined and should be guided by the resolution of all associated symptoms and findings.