Antifungal Prophylaxis in Pediatric Hematology/Oncology: New Choices & New Data

Abstract

A severe complication of the treatment of pediatric cancers is the development of an invasive fungal infection (IFI). The data to support antifungal prophylaxis in pediatric oncology patients derive primarily from adult patients, and thus the optimal agent to utilize is not clear. Fluconazole has been a standard option, but agents with antimold activity are now available, each with limitations. Pediatric dosing for voriconazole and posaconazole is uncertain and multiple drug interactions exist. The echinocandins are well-tolerated, but only available in intravenous form. Ultimately, studies demonstrating biologic risk factors for the development of IFI may lead to personalized prophylactic strategies.

INTRODUCTION

Children undergoing treatment for cancer are at an increased risk of developing an invasive fungal infection (IFI). The three major contributors to the development of an IFI are breakdown in natural barriers (such as indwelling catheter and mucositis), defects in cell-mediated immunity (lymphopenia from corticosteroids and other anti-T-cell cytotoxic agents), and deficient numbers of phagocytes (due to myelosuppressive chemotherapy) [1]. Having a single defect in the host defense system often is insufficient to allow for an opportunistic IFI, but with multiple defects, IFIs begin to emerge as a significant problem.

The data on IFI development, and their potential prevention, in immunocompromised hosts derive primarily from adult studies. However, children differ from adults in terms of both the types of IFI they develop or manifest, as well as their metabolism of antifungal agents. For example, invasive infections caused by Candida parapsilosis are more common, and Candida glabrata more rare, in children as compared to adults [2], and invasive aspergillosis (IA) may be more difficult to diagnose in children due to different radiologic findings [3]. Thus, extrapolating clinical decisions from adult trials may be problematic.

The European Organization for Research and Treatment of Cancer and Mycoses Study Group (EORTC/MSG) has established guidelines to standardize the definitions of proven, probable, and possible IFI [4]. However, in practice, the diagnosis of IFI is often difficult because of the lack of specific symptoms and the invasiveness of standard diagnostic tests. Significant attention has been focused on developing noninvasive assays such as galactomannan (GM) and β-D-glucan to diagnose IFI. GM is a polysaccharide cell-wall component that is released by Aspergillus during growth, and β-D-glucan is a cell wall polymer found in all fungi except Cryptococcus spp. and Zygomycetes. Although commercially available kits for detection of both GM and β-D-glucan are approved by the FDA for adults, the role of GM in diagnosing IFI in the pediatric population remains undefined [5] and data on β-D-glucan testing in pediatric patients are extremely limited. Until further research on these and other non-invasive tests is performed, the potential for early diagnosis of IFI in pediatric oncology patients remains elusive. Once an IFI develops, even in the modern era with newer antifungal agents, treatment success rates are suboptimal, especially for mold infections. For example, Burgos et al. [3] found that 53% of children diagnosed with IA died, and this case fatality rate rose to 78% in allogeneic hematopoietic cell transplant (HCT) recipients. It appears that for the present time, the optimal way to improve outcomes of IFIs in pediatric oncology patients is to prevent them from developing.

WHO DEVELOPS INVASIVE FUNGAL INFECTIONS?

Based on retrospective reports as well as toxicity data collected during therapeutic trials, there are several groups of pediatric patients that are at high-risk of developing an IFI: patients undergoing HCT (especially from an alternative allogeneic donor), patients receiving chemotherapy for acute myeloid leukemia (AML) or relapsed acute lymphoblastic leukemia (ALL), and patients with severe aplastic anemia (SAA) [3,6].

Although bacteria represent the most common infection following HCT, IFI account for a significant amount of post-transplant mortality. Several retrospective reports on the development of IFI in pediatric HCT patients have been published (Table I), with a 1-year incidence as high as 13–20% and a 58–83% mortality [7–10]. The most commonly identified invasive fungal organisms following HCT are Candida spp. and Aspergillus spp. Candida spp. are most commonly encountered during the neutropenic period immediately following HCT, while Aspergillus spp. have a bimodal distribution, with a first peak at a median of 16 days and the second at a median of 96 days post-allogeneic HCT [11,12]. Patients that are considered to be the highest risk for developing an IFI following HCT are those who undergo transplant from either an unrelated donor (including umbilical cord blood) or a partially matched related donor, for treatment of a malignancy, bone marrow failure syndrome, or congenital immunodeficiency [3,9,13–16].

TABLE I.

Incidence of IFI in Pediatric Oncology and HCT Patients

Study design	# of Patients (# of pediatric pts)	Disease/procedure	Type of prophy	IFI incidence	Reference
Retrospective	73 (73)	Auto or Allo HCT (37% alternative)	Variable	12.3% at 180 days	Hovi et al. [7]
Retrospective	221 (221)	Auto or Allo HCT (68% alternative)	Low-dose AMB	10.4% at 30 days 19.9% at 100 days	Benjamin et al. [8]
Retrospective	115 (115)	Allo HCT (47% alternative)	Fluconazole	8.7% at 30 days 12.5% at 365 days	Dvorak et al. [9]
Retrospective	97 (62)	UCBT	Fluconazole	12.4% at 30 days 22.6% at 100 days	Sadfar et al. [10]
UDMTT	404 (96)	URDHCT	Variable	31% at 2+ years	Van Burik et al. [12]
Prospective	2526 URDs (~9%)	URDHCT	Variable	7.7% at 365 days	Kontoyiannis et al. [16]
	474 MMRDs (~9%)	MMRD HCT		8.1% at 365 days
CCG 2961	492 (492)	New AML	Nonea	14–23% per phase	Sung et al. [17]
CCG 2891	335 (335)	New AML	Non-absorbablea	10–27% per phase	Sung et al. [18]
Prospective	18 (18)	New AML	Fluconazole or non-absorbable	29%	Groll et al. [19]
	7 (7)	Relapsed AML		28%
	97 (97)	New ALL		2%
	35 (35)	Relapsed ALL		9%
Retrospective	261 (261)	ALL	None	10%	Rosen et al. [20]
	117 (117)	AML		9%
Retrospective	425 (425)	New ALL	None	1%	Azfal et al. [21]
CHP-540	21 (21)	Relapsed ALL	Fluconazole	19%	Leahey et al. [22]
Retrospective	52 (?)	SAA	Fluconazole	36%	Torres et al. [23]

^aSome patients may have received systemic antifungals. UDMTT, unrelated donor marrow transplantation trial (prospective); CHP-540, Children’s Hospital Philadelphia Trial 540 (Prospective); CCG, Children’s Cancer Group; Auto, autologous; Allo, allogeneic; UCBT, umbilical cord blood transplant; URD, unrelated donor transplant; AML, acute myeloblastic leukemia; ALL, acute lymphoblastic leukemia; SAA, severe aplastic anemia; AMB, amphotericin B deoxycholate.

In children being treated for AML, several studies have demonstrated a high incidence (up to 29%) of IFI, both in newly diagnosed and relapsed patients (Table I) [17–20]. The high rates of molds and increased attributable mortality from them [18], suggests that this group of patients may benefit from antimold antifungal prophylaxis. Conversely, studies of patients with ALL (Table I) suggest that only those with relapsed disease have a high enough incidence of IFI to justify routine prophylaxis [19–22]. This finding makes sense from a biologic standpoint, in that patients with relapsed ALL have generally received years of lympholytic chemotherapy during which time they could have become colonized with fungal spores, and then are treated with more aggressive myelosuppressive chemotherapy for their relapsed disease. Finally, Patients with SAA are another susceptible group (Table I), where the percentage of deaths attributable to IFI is as high as 11% [23]. Even if patients with SAA do not initially show evidence of an IFI, they may have a higher rate of colonization during their neutropenic period, which puts them at higher risk for developing an IFI during their HCT [9].

Based on these data, it is reasonable to conclude that pediatric patients receiving treatment for AML and those undergoing allogeneic HCT would potentially benefit from antifungal prophylaxis. Additionally, patients undergoing autologous HCT, receiving therapy for relapsed ALL, and those with SAA should also be considered for antifungal prophylaxis. The risk of IFI in patients receiving treatment for newly diagnosed ALL or a solid tumor is not high enough risk to justify routine use of prophylactic antifungals.

Before we commit patients to antifungal prophylaxis, a key question is whether antifungal prophylaxis is beneficial? Robenshtok et al. performed a meta-analysis of 64 antifungal prophylaxis trials in an attempt to answer this question [24]. They demonstrated that for patients undergoing allogeneic HCT, antifungal prophylaxis significantly decreased the risk of IFI and all-cause mortality, while for patients undergoing autologous HCT, the diminution in IFI was not statistically significant, but all-cause mortality was still lower in the prophylaxis group. Conversely, in patients with acute leukemia, the risk of IFI development was lower with antifungal prophylaxis, but this did not result in a statistical improvement in all-cause mortality. Certainly, these data are supportive of antifungal prophylaxis in the aforementioned high-risk groups. Unfortunately, only five of the analyzed trials included pediatric patients and thus caution should be used when applying these data.

WHAT IS THE BEST PROPHYLACTIC ANTIFUNGAL AGENT IN CHILDREN?

Several published randomized prospective trials comparing antifungal agents have included pediatric patients, but usually only as young as 12 years of age [25–27] with a few exceptions that included a small number of younger patients [28,29], and pediatric patients have generally not been separately analyzed (Table II). Therefore, conclusions about the optimal prophylactic agent in pediatric oncology or HCT patients are based almost exclusively upon extrapolation from adult data.

TABLE II.

Selected Antifungal Prophylaxis Trials

Prophylaxis	Design	# of Patients (# of pediatric)	Disease/procedure	Control outcome	Intervention outcome	Reference
Fluconazole	DB, PC, MC	356 (?)a	Auto & Allo HCT	16% at 50 days (placebo)	3% at 50 days	Goodman et al. [25]
Fluconazole	DB, PC, SC	300 (?)a	Auto & Allo HCT	18% at 75 days (placebo)	7% at 75 days	Slavin et al. [26]
Amphotericin B	OL, MC	355 (0)	Auto & Allo HCT	9% (fluconazole)	14%	Wolff et al. [30]
LFAB	OL Pilot	57 (57)	Allo HCT	NA	0% at 100 days	Roman et al. [31]
Itraconazole	OL, SC	304 (5)a	Allo HCT	15% (fluconazole)	7%	Marr et al. [37]
Voriconazole	DB, MC	600 (51)	Allo HCT	11% (fluconazole) at 6 months	7% at 6 months	Wingard et al. [28]
Posaconazole	OL, MC	602 (?)a	AML/MDS	11% (flucon or itra) at 100 days	5% at 100 days	Cornely et al. [27]
Micafungin	DB, MC	882 (84)	Auto & Allo HCT	1.6% (fluconazole) at 70 days	2.4% at 70 days	van Burik et al. [29]
Caspofungin	OL, SC	200 (0)	AML/MDS	6% (itraconazole)	6%	Mattiuzzi et al. [50]
Caspofungin	Retrospective	123 (0)	Auto & Allo HCT	NA	7.3% at 100 days	Chou et al. [51]

^aOnly patients >12 years of age were eligible; DB, Double-blind; PC, Placebo-controlled; MC, Multi-center; OL, Open-label; SC, Single-center; NA, Not Applicable; AML, Acute Myeloblastic Leukemia; MDS, Myelodysplastic Syndrome.

Currently, the most commonly recommended agent for anti-fungal prophylaxis in high-risk children has been fluconazole. This is based on two placebo-controlled trials from the early 1990s, performed in patients >12 years of age undergoing autologous or allogeneic HCT [25,26]. Long-term follow-up of allogeneic HCT patients on fluconazole supports a survival benefit (mortality 20% in the fluconazole arm vs. 35% in the placebo arm; P = 0.004) postulated to be secondary to less severe graft versus host disease of the gut in the fluconazole recipients from decreased intestinal antigenic stimulation [13].

Although it reduced the risk IFI relative to placebo, fluconazole lacks activity against Aspergillus spp, which is the second most common cause of IFI in these patients. Given this lack of antimold activity, several trials have compared it to mold-active agents in hopes of further decreasing rates of IFI. The first of these trials compared fluconazole to low-dose conventional deoxycholate amphotericin B deoxycholate (D-AMB) [30]. However, D-AMB did not show improvement over fluconazole and resulted in a higher adverse event rate. With the advent of liposomal amphotericin B (LFAB) there was renewed interest in prophylaxis with an amphotericin B product. Several trials, including one in children [31], have evaluated LFAB (often given only three times per week) for antifungal prophylaxis in HCT and acute leukemia patients [32–34]. However, like D-AMB, LFAB has not been shown to be superior to fluconazole and typically demonstrated increased side-effects.

Extended-spectrum azoles such as itraconazole, voriconazole, and posaconazole do possess anti-Aspergillus activity [35]; however, there are clear limitations to each as a potential prophylactic agent. Several trials of itraconazole versus fluconazole have been performed, and a meta-analysis showed significantly less IFI [36], but because of its side effects, greater drug interactions, and poor tolerability [37], itraconazole prophylaxis has generally been abandoned in children. The results of a multi-center, double-blinded trial showed that voriconazole was not superior to fluconazole in the prevention of IFI, though the safety profile was similar [28]. Given voriconazole’s broader spectrum of activity, this result was surprising, but may have been due to an incomplete understanding of the complex pharmacokinetics of voriconazole. In adults and children over the age of 12 years (the subjects of most trials to date), voriconazole has nonlinear pharmacokinetics with relatively well-established dosing regimens. Even in adults, however, recent studies have questioned standard dosing regimens, and have proposed dosing based on serum drug levels [38–40], though the optimal serum voriconazole level is still uncertain. Part of this variability may be due to allelic polymorphisms of the gene encoding for CYP2C19, which can result in an increase or decrease in voriconazole metabolism [40]. In children, the situation is further complicated by linear kinetics. In children between the ages of 2 and 12 years, the optimal dose may be 7 mg/kg twice daily [41,42], while in children <2 years of age, it may be as high as 8.5 mg/kg/dose twice daily [43]. Even more recent data have led to a proposed maintenance dosing of 8 mg/kg twice daily for all children less than 12 years of age, and for those 12–14 years of age weighing less than 50 kg [44]. Currently there is no universally accepted approach to dosing and/or monitoring of serum levels.

Voriconazole also has significant drug interactions with commonly used agents in a pediatric oncology or HCT population. Voriconazole is a substrate of CYP2CP (major), 2C19 (major) and 3A4 (minor) and an inhibitor of 2C9 (moderate), 2C19 (weak), and 3A4 (moderate) [45]. Proton pump inhibitors increase voriconazole levels, while voriconazole increases serum levels and/or toxicity of corticosteroids, vincristine, imatinib, bortezomib, ironotecan, and many other medications [45]. The FDA label reports that concomitant use of voriconazole can cause a 1.7- to 3-fold increase in cyclosporine or tacrolimus levels, and recommends that the dosing of cyclosporine be decreased by 50% and the dosing of tacrolimus be decreased by 66% of the normal dose. A recent study demonstrated in a cohort of 27 adult HCT patients receiving voriconazole, 100% of patients required multiple tacrolimus dose reductions to achieve a safe target level [46]. Furthermore, the use of voriconazole with sirolimus is officially contraindicated, and when its use has been reported, investigators have recommended dropping the levels of sirolimus to 90% of original dosing at the time of initiation of voriconazole [47].

Posaconazole is a triazole with broad coverage of most fungi, including zygomycetes [35]. In a trial in adult patients with neutropenia, posaconazole prophylaxis was superior to fluconazole or itraconazole, but was also associated with an increased risk of serious adverse events [27]. Factors limiting the use of posaconazole in children include the lack of pharmacokinetic data in children less than 13 years of age, the lack of an intravenous formulation (since these children typically have nausea, mucositis, and poor oral intake during the at-risk period), and unreliable absorption in the setting of limited oral intake. Finally, posaconazole shares many of the same enzymatic pathways, and therefore drug interactions, as voriconazole.

The echinocandins are a novel class of antifungal agents that target β-(1,3)-D-glucan synthase and interrupt biosynthesis of the glucan polymers that make up fungal cell walls. Because mammalian cells do not possess cell walls, echinocandin administration to human patients has resulted in minimal toxicity. Echinocandins possess fungicidal activity against Candida spp. (including Candida krusei and Candida glabrata, which possess significant degrees of fluconazole resistance) and Pneumocystis jiroveci, as well as fungistatic activity against Aspergillus spp. [35]. The echinocandins may be superior to fluconazole [48] or amphotericin B [49] for treatment of invasive candidiasis which, when combined with their anti-Aspergillus activity and excellent safety profile, makes them an attractive option for antifungal prophylaxis. In a prophylactic antifungal trial, micafungin demonstrated reduced need for empiric antifungal therapy and an improved safety profile compared to fluconazole [29]. However, the number of pediatric subjects enrolled was small (n = 84) and a reduction in the incidence of proven or probable IFI was not demonstrated. The lack of impact on IFIs may have been because the incidence of breakthrough IFIs in both groups was very low, likely due to the inclusion of low-risk patients (46% autologous HCT recipients) and very few patients undergoing umbilical cord blood transplant (UCBT; n = 30). Caspofungin has been shown to be at least equivalent to itraconazole in the setting of antifungal prophylaxis [50], with little caspofungin-related adverse events [51]. The major disadvantages to widespread echinocandin use are cost and the lack of an available oral formulation.

The lack of pediatric trial data on antifungal prophylaxis and the limitations inherent to each antifungal agent make it challenging to define the optimal antifungal prophylaxis for pediatric patients. In 2009, a survey of both the Children’s Oncology Group (COG) and BFM Consortia was performed to determine local antifungal prophylaxis choices for patients with AML. Interestingly, only 77% of COG centers (n = 180) utilized antifungal prophylaxis in their patients with AML, compared to 91% of BFM centers (n = 47). Of those COG institutions that routinely give antifungal prophylaxis, 82% utilized fluconazole, 21% voriconazole, and 2% each for posaconazole or a LFAB (with some centers using more than one agent). There is less uniformity within the BFM, where centers were relatively equally likely to use fluconazole (28%), voriconazole (24%), posaconazole (13%), or a LFAB (24%) [52]. In 2011, a similar survey was performed in the 80 COG HCT Centers specifically asking about prophylactic choices for alternative donor (unrelated or mismatched related) HCT recipients. The survey found that for unrelated donor HCT recipients (including UCBT), 56% of centers continue to use fluconazole, while 28% use voriconazole, 11% an echinocandin, and 5% administer a LFAB. The data were very similar for patients undergoing mismatch related donor HCT (57%, 27%, 8%, and 8%, respectively).

These results indicate that the profound lack of data for this patient population has led to a clinical situation where there is no clear agent of choice for patients at high risk of developing an IFI. Because of this, in April 2011 the COG initiated a randomized open-label trial of caspofungin compared to fluconazole to prevent IFI in children undergoing chemotherapy for AML (ACCL0933). Furthermore, an upcoming second COG trial (ACCL1131) will ask a similar question in pediatric recipients of alternative donor HCT, comparing caspofungin to center-choice of fluconazole or voriconazole (due to high rates of voriconazole usage).

INVASIVE FUNGAL INFECTIONS: BEYOND TRADITIONAL RISK FACTORS?

As noted previously a number of therapy-induced alterations of host defenses have been identified as risk factors for IFI. However: why does one child develop an IFI, while another similarly age child with the same disease, being treated at the same institution using the same type of chemotherapy or immunosuppressive agents, does not develop an IFI? There is considerable evidence emerging that there is a genetic component to the susceptibility and outcome of IFI in immunocompromised populations. In allogeneic HCT recipients, several polymorphisms in both host and donor genes appear to significantly predispose patients to IFI (Table III) [53–59]. Future investigation will likely uncover additional polymorphisms that place immunocompromised hosts at increased risk of IFI. As more details on genetic risk factors are unveiled and validated, risk stratification will be improved beyond the current system that only utilizes traditional IFI risk factors. Furthermore, although all of these studies have been performed in allogeneic HCT patients, there is no biologic reason to suspect that these polymorphisms will not also play a role in the development of IFI during treatment of AML, relapsed ALL, SAA, and potentially even other “low-risk” diseases.

TABLE III.

Genetic Risk Factors for the Development of IFI Following Allogeneic HCT

Infection	Gene	Source	# of HCTs	Hypothetical mechanism	Reference
IA	TLR1 & TLR6	Host	127	Decreased recognition by phagocytes	Kesh et al. [53]
IA	IL-10 promoter	Host	105	Less production of IL-10	Seo et al. [54]
IA	Plasminogen	Host	236	Increased tissue damage & invasion	Zass et al. [55]
IA	TLR4	Donor	366	Decreased recognition by phagocytes	Bochud et al. [56]
IA	Chemokine Ligand 10	Donor	139	Less response to IFN-y, so less TH1 cells	Mezger et al. [57]
IA	Dectin-1	Both	205	Less production of IFN-y & IL-10	Cunha et al. [58]
IFI	MBL	Donor	106	Decreased complement fixation	Granell et al. [59]
IFI	MASP2	Host	106	Decreased complement fixation	Granell et al. [59]

IA, invasive aspergillosis; IFI, invasive fungal infection; IL, interleukin; IFN, interferon, TH1, T helper 1.

CONCLUSIONS

In summary, administration of antifungal prophylaxis to patients receiving chemotherapy for AML or relapsed ALL, patients with SAA, and patients undergoing any form of HCT should be considered standard of care. At present, fluconazole remains the first line therapy, although the significance of invasive mold infections in these patients, especially Aspergillus spp., suggest that an agent with antimold activity and a favorable safety profile may be superior. Although used by some centers treating children with AML and those undergoing allogeneic HCT, neither LFAB nor the triazoles (voriconazole and posaconazole) are conclusively superior prophylactic agents compared to fluconazole based on adult efficacy and safety data, and lack of optimal pediatric dosing. The echinocandin class of antifungals may represent a balance between broad coverage of the most commonly encountered organisms coupled with a favorable safety profile. The current COG trials comparing an echinocandin to azole therapy in AML and high-risk allogeneic HCT populations should help to determine their relative efficacy in these patient populations. These data may then potentially be extrapolated to other high-risk groups, such as relapsed ALL or SAA, though clearly the definitive answer could only be provided by a trial in the specific at-risk population.

Furthermore, additional future research should undercover genetic risk factors that predispose to the development of an IFI. This could then be used to more clearly define risk-stratification groups (analogous to that used for ALL therapy) in which only the standard-risk patients would receive antifungal prophylaxis, while, low-risk patients could be closely monitored in a preemptive fashion with initiation of antifungal therapy at the first evidence of IFI (e.g., positive serum biomarkers). Finally, the high-risk patients could be the focus for future prophylaxis trials, either of novel antifungal agents, or for potentially synergistic combination therapy, such as with a triazole and an echinocandin [60]. Such a therapeutic strategy could reduce overall antifungal exposures with resultant reduction in adverse events, decreased cost, and decreased pressure for evolution of resistant organisms.