Treatment of invasive fungal diseases in cancer patients—Revised 2019 Recommendations of the Infectious Diseases Working Party (AGIHO) of the German Society of Hematology and Oncology (DGHO)

Invasive fungal diseases remain a major cause of morbidity and mortality in cancer patients undergoing intensive cytotoxic therapy. The choice of the most appropriate antifungal treatment (AFT) depends on the fungal species suspected or identified, the patient's risk factors (eg length and depth of granulocytopenia) and the expected side effects.


| BACKG ROU N D
In cancer patients, invasive fungal diseases (IFDs) remain an important complication still causing high mortality and morbidity.
Chemotherapy or transplantation procedures are often delayed or postponed in patients with IFD which might lead to poor overall survival, in particular after stem cell transplantation. Adherence to guidelines was found to be suboptimal in the past, but adherence to guidelines may lead to a higher response rate to first-line antifungal treatment (AFT) of invasive aspergillosis in leukaemic patients. 1 In recent years, recommended treatment strategies were gradually moving away from solely empirical therapy of possible IFD towards pre-emptive therapy of probable IFD. AFT of IFDs in cancer patients may include not only antifungal agents but non-drug treatment as well. Furthermore, new antifungal agents have been

| Empirical vs pre-emptive antifungal therapy
The time point of initiation of antifungal therapy (AFT) in granulocytopenic high-risk patients with fever and prolonged granulocytopenia is critical. Current guidelines 9 recommend starting empirical systemic mold-active AFT in this patient cohort in case of persistent fever of unknown origin (FUO) after 4-6 days of broad-spectrum anti-pseudomonal beta-lactams. In a recent meta-analysis, 10 this empirical antifungal strategy showed a high efficacy, favouring echinocandins as the preferable class of agents. However, it has drawbacks including the risk of side effects, drug-drug interactions, emergence of resistant fungal pathogens and costs, and another meta-analysis supports the use of pre-emptive antifungal therapy by showing non-inferiority and substantial resource reduction in comparison with the empirical approach. 11 In parallel, diagnostic efforts to identify a source of infection, for example, pulmonary infiltrates suggestive of invasive mold infection 8 and serial testing for fungal biomarkers such as Aspergillus galactomannan alone or in combination with molecular targets by using PCR assays [12][13][14][15] have been strongly advocated in addition to repeated blood cultures and physical examinations. 9 For AFT implementation, this pre-emptive or 'diagnostic-driven' therapy (ie the diagnostic work-up shows suspicious findings before initiation of antifungal treatment) has been compared with empirical ('fever-driven') AFT. 14,[16][17][18][19] While overall and infection-related mortality did not show statistically significant differences, the rate of proven or probable invasive fungal disease has been substantially higher in patients not treated empirically. As a result, the routine use of the diagnostic-driven approach cannot be recommended as long as the current diagnostic tools lack sensitivity and/or specificity and thresholds triggering AFT are not clearly defined (BII). Furthermore, treatment delay might enhance mortality in this patient population. Efforts to further reduce the risk of IFD by TA B L E 1 Grading of recommendations, adopted from the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) [3][4][5][6] Category, grade Definition Empirical and pre-emptive antifungal treatment is not mutually exclusive. [21][22][23] In granulocytopenic high-risk patients with FUO, em-

| Treatment of invasive aspergillosis
Acute invasive pulmonary aspergillosis (IPA) is the most frequent manifestation of systemic/invasive aspergillosis (IA) in granulocytopenic patients 25 with a fatality rate that ranges from 30% to 60%. [26][27][28] Early treatment at first signs of infection is mandatory and improves the chance of survival (AIII). 29 See documented in many body sites including brain parenchyma. [56][57][58] However, a large variability in trough plasma levels has been observed. 59,60 Studies demonstrated a positive correlation between plasma levels, clinical efficacy and toxicity. Plasma concentrations of >1 mg/L were found to be correlated with response to therapy.
However, plasma levels >5.5 mg/L were associated with neurotoxicity. 61 In contrast, in lung transplant recipients a cut-off for toxicity was not identified. 61  When used for salvage treatment, caspofungin resulted in a response rate of 45%-49% in two non-comparative studies of patients with invasive aspergillosis and failure of or intolerability to standard antifungal therapy. 75,76 A case collection of 118 patients demonstrated a response rate of 61%. 77 In the CAN-DO study, 45 from 81 patients responded to caspofungin treatment. 70 Caspofungin is recommended for salvage therapy (BII).
Micafungin: Micafungin has been investigated mostly in salvage therapy studies and retrospective analyses as mono-and particularly combination therapy which resulted in efficacy rates of about 25%-36%. 78,79 Anidulafungin: Anidulafungin as monotherapy for treatment of IA has not been studied properly to allow inclusion of this drug into the therapy algorithms as first-line (or even salvage) therapy of IA. 80,81 In combination with voriconazole, an additive efficacy and reduced mortality rate at six weeks has been observed as compared to voriconazole monotherapy (see chapter combination therapy). 82 TA B L E 3 Recommendations for specific invasive fungal diseases incl. therapeutic drug monitoring. Invasive/systemic aspergillosis-(a) first-line therapy and (b) second-line/salvage therapy drugs. Renal toxicity was less frequent with ABCD but infusion-related toxicity was higher. This profile led to the recommendation against the use of ABCD (DI). In addition, the drug is no more available in Europe.

Liposomal amphotericin B (L-AmB):
There are several liposomal AmB products available, but only one licensed L-AmB (AmBisome ® ) is available in Europe and North America. 84 Several non-comparative studies with L-AmB for second-line therapy exist from the early 1990s which included only smaller numbers of patients and resulted in response rates of 50%-70%. 85 In a pooled efficacy analysis, L-AmB therapy resulted in a response rate of 47% for the treatment of invasive aspergillosis. 86 In a randomised study, L-AmB was equally efficacious compared to D-AmB in the first-line therapy of invasive mycosis, 87 but the study was not restricted to patients with IA. The efficacy of L-AmB vs ABLC in the first-line therapy has been compared in an analysis of eight open-label studies with more than 1000 patients resulting in a response rate of 61% vs 46% favouring L-AmB over ABLC. 88 A retrospective study in 158 consecutive patients with mainly acute leukaemia or allogeneic stem cell transplantation receiving L-AmB or ABLC for invasive aspergillosis resulted in a poor outcome of both groups (12%). 89 ABLC was associated with significantly higher nephrotoxicity rates compared to L-AmB. 89 The studied dosages of L-AmB for treatment of invasive aspergillosis are 1-10 mg/kg/d (manufacturer recommendation: 1-5 mg/ kg). 30,87,90 A randomised study comparing L-AmB 4 mg/kg vs 1 mg/ kg resulted in similar efficacy rates, but survival at day 14 and response in patients with proven aspergillosis was higher in the 4 mg/ kg arm. 90 A randomised comparison of L-AmB 3 mg/kg vs 10 mg/kg (mainly cancer patients) in first-line therapy of invasive aspergillosis showed equal efficacy but an increased toxicity with the higher dosage. 30 The response rate was high and comparable to voriconazole.
In a recent pharmacokinetic study with L-AmB in obese individuals (>100 kg), it was calculated that a fixed dose of 300 mg L-AmB may be an alternative instead of 3 mg/kg L-AmB. 91 The AGIHO recommends L-AmB (3 mg/kg) for the first-line treatment of IA with lesser strength than isavuconazole or voriconazole (AII), since all available trials did not compare L-AmB with a standard treatment. L-AmB may be also used as second-line treatment (BII). 92 Amphotericin B deoxycholate (D-AmB): Intravenous therapy with D-AmB had been the therapeutic gold standard for IA with response rates of 30 (−50)% for many years in the past. 93 Maximum tolerable daily dosages of up to 1.5 mg/kg have been recommended.
Comparative clinical studies on dose regimens are, however, not available. Due to its high toxicity and inferiority compared to voriconazole in a randomised controlled study, 31 we strongly discourage the use D-AmB (DI).

Combination therapy
The benefit of combination of D-AmB plus 5-flucytosine has not been substantiated by appropriate clinical trials. 94

Salvage therapy
Response to antifungal therapy in patients with invasive mold disease may be defined either as success (complete or partial) or failure (stable, progression or death). 100 It is not clear, whether and when patient with a stable response (minor or no improvement of signs and symptoms or persistent isolation of moulds) should receive a salvage therapy. However, when progression of disease is evident (with worsening of signs and symptoms plus new sites of disease or radiological worsening) salvage therapy is indicated.
In general, radiological improvement may be not observed unless a minimum of 7-14 days of full-dose treatment is given. It was observed that despite administration of effective antifungal treatment, the median volume of lesions increased fourfold during the first week before these lesions stabilised or improved during the second week. 101 Therefore, in a clinical stable situation a reliable clinical response may not be assessed before 10-14 days of adequate therapy (BIII).
Apart from evident failure due to intrinsic resistance of the pathogen (eg A terreus to AmB), lack of adequate drug levels (see chapter therapeutic drug monitoring), intolerance or severe organ toxicity, non-response of IA to an established antifungal therapy should be stated with caution. 100

Duration of antifungal treatment
Generally, the antifungal therapy should be continued during the

| Other manifestations
Invasive sinus aspergillosis: Aspergillus sinusitis was described in individuals with acute leukaemia or after allogeneic stem cell transplantation. 112 The IFD is primarily caused by A flavus or A fumigatus. 113 Frequently, additional surgical debridement is required (BII) (see chapter interventional strategies). Overall, Aspergillus sinusitis has been associated with a mortality rate ranging from 26% to 66% while treated with conventional AmB. 114 Therapy recommendations do not differ from pulmonary manifestations (see Table 2).
Aspergillosis of the CNS: Aspergillus spp. rarely cause meningitis or micro-abscesses of the brain, but macro-abscesses-especially in severely immunocompromised patients-are most often caused by A fumigatus (followed by other moulds such as Mucor spp.). In the majority of patients with cerebral IA, the CNS is invaded by haematogenous spread from primary sites of infection such as the lungs. Acute leukaemia is the most common underlying disease. 115 Patients with aspergillosis within the CNS typically present with focal neurological signs such as pareses or seizures. Overall mortality is still high reaching 69% (with IFD-attributable mortality 33%) in a recent study from Italy. 115 Comparable studies regarding drug treatment of CNS aspergillosis do not exist, but D-AmB was found to be not effective. 116 Due to its good penetration into the cerebrospinal fluid and brain tissue, voriconazole is recommended for primary treatment and has shown a survival rate of Mucor, were reported to be alive at day 84. 119 These data suggest that isavuconazole is similar effective to voriconazole but is active against Mucor spp. as well (AII).
Alternatively, L-AmB might be administered in case of contraindication, intolerance or poor response to voriconazole (BIII). According to data from animal studies, significantly enhanced activity was found with the combination therapy of L-AmB plus voriconazole. 120 The role of echinocandins has not been fully explored other than in case reports. 121 A retrospective study of 81 patients with CNS aspergillosis resulted in significantly better survival in patients undergoing surgery. 117 Therefore, surgical resection of singular lesions is recommended together with systemic AFT (AII).

| Treatment of invasive candidosis
In the past, the most common cause of IFD in cancer patients was yeast pathogens, in particular Candida albicans, followed by non-  127,128 Prospective trials in granulocytopenic patients will be probably never performed due to small numbers of patients. Therefore, recommendations are adapted to those in non-granulocytopenic cancer patients. The role of catheter removal in granulocytopenic patients is particularly controversial as the gastrointestinal mucosa, damaged by cytotoxic chemotherapy, is thought to be the main port of entry for yeasts into the bloodstream. 129 However, as the central venous line might be colonised, its removal is recommended in these patients as well as in non-granulocytopenic patients by the AGIHO (AII). 6

Echinocandins
Echinocandins were not prospectively studied in granulocytopenic patients. In a pooled, post hoc analysis of phase 3 trials, the over- more rapid eradication of yeasts from bloodstream compared to fluconazole alone. 143 The combination therapy with D-AmB plus flucytosine has been advocated in earlier times in particular for children with acute myeloid leukaemia and IFDs, 144 but was never studied properly in a subsequent trial. 145,146 In addition, the use of flucytosine needs regular monitoring of plasma levels to avoid toxicity (eg haematotoxicity). 56,147 In summary, in adult patients with cancer or haematological malignancies there are only limited data which support a recommendation of combination therapies for invasive Candida infections (CIII).

Salvage therapy
Data on second-line therapy in cancer patients, in particular during granulocytopenia, are limited to case reports, and specific recommendations cannot be given.

Duration of antifungal therapy
Duration of treatment in non-granulocytopenic patients is recommended for at least 14 days after the first negative blood culture and resolution of signs and symptoms of candidemia (BI), 6,148 but should be adapted in case of organ manifestations. In individuals who remain granulocytopenic but do have negative blood cultures should be evaluated for resolution of all signs and symptoms of IC before antifungal therapy is stopped (CIII). 6 teomyelitis. 149 This entity was mostly reported before the use of azole prophylaxis in leukaemia patients and HSCT and appears rare today. 150 Echinocandins and L-AmB may be recommended as initial antifungal treatment (AIt).

| Acute disseminated candidosis
In all cancer patients, fundoscopy and abdominal ultrasound (liver, spleen, kidneys) should be performed (during and after recovery from granulocytopenia) to exclude chronic disseminated infection/ hepato-splenic candidosis that may not be associated with clinical symptoms other than fever (BIII).

| Management of intravenous lines
Intravenous lines should be removed in cancer patients at initiation of antifungal therapy whenever feasible to reduce IFD-related mor-

| Chronic disseminated candidosis
If fever persists after neutrophil recovery, chronic disseminated candidosis (CDC; hepatosplenic candidosis) may be considered in hae-  167,168 (CIII). In stable patients, intravenous therapy may be switched to oral medication (step down strategy; eg ≥ 5 days iv AFT) (BIIt). This strategy has not been studied in CDC so far, but is regarded as safe and effective in patients with candidemia (see above). 169

| Treatment of mucormycosis
Mucormycosis is an emerging invasive fungal infection in patients with haematological malignancies and allogeneic stem cell transplantation. 179 In granulocytopenic patients, it usually involves the lung and causes high mortality rates. The clinical presentation is difficult to distinguish from invasive pulmonary aspergillosis. 180,181 A socalled reversed halo sign has been described on computed tomography scans, but is not entirely specific for mucormycosis. 182 10 mg/kg/d iv) for mucormycosis. [190][191][192][193] Posaconazole has been studied primarily for second-line or salvage therapy in small case series but not for first-line treatment. 48,[194][195][196] Isavuconazole has been studied in a single-arm open-label trial (VITAL study) in 37 patients for a median of 84 days. 187 Day-42 crude all-cause mortality was 33% and efficacy was found similar to amphotericin B. See Table 5a,b.
In a small series, antifungal combination therapy has been reported. Posaconazole plus L-AmB (either 3 mg/kg or 5 mg/kg) has been successfully used in 27 patients. 197 L-AmB has been successfully combined with caspofungin for rhino-orbital-cerebral diseases in mostly diabetic non-cancer patients 198   are not active against Cryptococcus spp. In vitro, and breakthrough disseminated cryptococcal disease has been reported. 234,235 Therefore, echinocandins should not be used for treatment of cryptococcosis (DI).

| Treatment of fusariosis
Invasive fusariosis is a severe sporadic mold infection affecting mainly granulocytopenic patients. 236,237 It is associated with a very high mortality rate ranging from 50% to 80%. 236  Following infections with Trichosporon spp., appearance of hepatic and splenic lesions with the recovery from granulocytopenia has been described similar to IRIS in disseminated candidosis. 257 Treatment of invasive trichosporonosis remains a challenge, and no data from prospective trials are available. High fatality rates were reported from granulocytopenic patients with acute leukaemia (crude mortality up to 77%). 253,256 Response to D-AmB was reported to be poor in 55 patients from Italy (response in 24% of patients). 253 Trichosporon asahii isolates exhibit often high MICs to Amphotericin B in vitro. Therefore, D-AmB cannot be recommended for first-line monotherapy (DIII). Echinocandins exhibit no in vitro activity against Trichosporon spp. and should not be used (DIII). 252 Breakthrough trichosporonosis has been repeatedly reported in patients with haematological malignancies receiving micafungin but also rarely on D-AmB and azole therapy (eg during prophylaxis with itraconazole or posaconazole). 256,[258][259][260][261] Response to AFT and survival was best when patients receive azole therapy (fluconazole, voriconazole) (CIII). 262 The most common pathogens are Lomentospora prolificans (formerly Scedosporium prolificans) and Scedosporium apiospermum (formerly Pseudallescheria boydii). 273 While L prolificans typically occur in immunocompromised patients, S apiospermum is often reported in immunocompetent individuals after near-drowning. 241 Systemic infections with Scedosporium species are often refractory to treatment as these pathogens are highly resistant to most available antifungal agents. [278][279][280] Patients with disseminated L prolificans infection often have positive blood culture (up to 70%). 281 However, most blood cultures become positive shortly before death and antifungal therapy often failed in the terminally ill patient. 241,275 Malignancy, fungemia, CNS and lung involvement predicted a adverse outcome. 281 According to a multivariate analysis of 162 cases, survival was independently associated with surgical excision and recovery from aplasia but not from antifungal therapy (not specified). 275 No treatment data from randomised trials exist for any patient group, and available information about treatment outcomes is available only from case reports and case collections. According to a large registry with 264 cases, patients treated with voriconazole had a better outcome compared to treatment with amphotericin B

| Treatment of trichosporonosis
formulations. 281 Voriconazole has better in vitro activity against S apiospermum as compared to L prolificans and is regarded as drug of choice for disseminated scedosporidiosis (BII). 280,282 Other azoles such as posaconazole or isavuconazole do have similar in vitro activity against S apiospermum compared to voriconazole and might serve as alternatives. 283 The clinical activity of D-AmB and/ or L-AmB is unclear in granulocytopenic patients and cannot be currently recommended (DIII). In a small case series of 25 patients, the majority of survivors received a combination therapy consisting of L-AmB and a triazole (voriconazole, posaconazole). 277 Given these data, the treatment recommendation for S apiospermum infection is an azole such as voriconazole or posaconazole plus surgical debridement (BIII).
Results from in vitro testing found a synergistic effect in the combination of voriconazole plus terbinafine against L prolificans. 284 This in vitro effect was translated into clinical practice, showing a response in some patients with clearance of disseminated L prolificans infection. [285][286][287][288][289] The combination of azoles plus echinocandin revealed conflicting results and clinical data are lacking (DIII). In an in vitro study, the azole/echinocandin (micafungin) combination did not show a better in vitro activity when compared to voriconazole or posaconazole monotherapy. 290 However, in an animal model micafungin combined with voriconazole or amphotericin B was effective in reducing fungal burden and prolonging survival. 291 As a potential option serves the combination of voriconazole with miltefosine which showed synergy against L prolificans isolates in vitro and was successfully used in a child with refractory L prolificans osteomyelitis. 292,293 Recommendation: Taken together, voriconazole plus terbinafine appears to be the best currently available treatment for invasive scedosporidiosis in patients with haematological malignancies (CIII).
For more detailed information the reader is referred to the detailed ESCMID/ECMM guideline. 247

| Therapeutic drug monitoring of antifungal agents
Pharmacokinetic properties of antifungal agents vary substantially, and bioavailability might have an impact on clinical efficacy. For flucytosine with its known association of plasma concentrations with toxicity, therapeutic drug monitoring (TDM) has broadly been established. 147,294,295 For flucytosine, a plasma target concentration of 30-80 mg/mL two hours after application is recommended (BIIt). For azole antifungal drugs, therapeutic drug monitoring (TDM) has been frequently studied to guide AFT, especially to avoid toxicity. 59,63,296,297 Voriconazole plasma concentrations show a broad range of intraand interindividual variation. 60,63,298,299 In recipients of an allogeneic HSCT, exposure and clearance of voriconazole are similar to those of healthy volunteers though there was high intra-and interindividual variation in drug exposures. 300 This is caused by potential drug interactions due to metabolisation through the cytochrome P450 system, altered biodegradation due to genetic variations of isoenzyme CYP2C19 and other factors including food, co-medication and absorption. 60,297,298,301 In consequence, voriconazole plasma concentrations cannot be predicted by dosage. 60,63,302 According to a recent meta-analysis, patients with therapeutic voriconazole serum concentrations were twice as likely to achieve successful outcomes. 60,63,298 An increased rate of adverse events with high plasma concentrations (usually above 5.0-5.5 mg/L) has been reported. 57,60,298 With regard to efficacy, the serum level should exceed 1-2 mg/L, while one study found a significantly higher treatment failure rate when voriconazole levels were <1.7 mg/L as compared to >1.7 mg/l. 57,302 Multiple regression analyses of voriconazole concentration identi-

| Drug instillation
For treatment of refractory abscesses, cavities (eg in the lung) or severe haemoptysis from pulmonary aspergilloma in which surgical intervention is not feasible, a drainage (in particular for fungal empyema) as well as a local drug instillation may be considered. 321,322 Here, antifungal preparations (commonly containing AmB preparation or azoles, eg voriconazole) have been used (CIII). [323][324][325] No change to previous AGIHO recommendations.

| Embolisation
Embolisation may be considered in the case of large pulmonary infiltrates where the occurrence of severe haemoptysis due to vessel erosion is likely, including the development of aneurysms. 326 The use of CT perfusion (by computed tomography pulmonary angiography) in case of suspicion of angioinvasive pulmonary mycosis proved beneficial and could discriminate from other pulmonary infections. 327 In this case, the bronchial and pulmonary vessels may be embolised (CIII). No change to previous AGIHO recommendations. Another case-controlled study in patients with candidemia showed an equal short-term survival rate, but the group with granulocyte transfusions had higher risk factors which may be interpreted as a benefit of this option. 333 In 31 patients with invasive fungal infection (17 possible infections) undergoing granulocyte transfusions, 78% survived. 334 A randomised study with prophylactic granulocyte transfusion three times a week in patients with granulocytopenic fever and pulmonary infiltrates or a history of proven IFD failed to confirm the benefit of this procedure. 335 Currently, a clear benefit of granulocyte transfusions in IFDs has not been proved. 336 However, it might be considered as a treatment option in selected patients (CIII).