Management of pulmonary aspergillosis in children: a systematic review

Aspergillus spp. is a ubiquitous, slow-growing mold that commonly colonises the respiratory tract. Depending on the host immune status and lung structure, it can manifest as different clinical entities, such as aspergilloma, allergic bronchopulmonary aspergillosis (ABPA), allergic sinusitis, invasive aspergillosis, chronic pulmonary aspergillosis [1]. Allergic sinusitis and ABPA are allergic responses to colonisation by Aspergillus spp. occurring in immunocompetent patients. In particular, ABPA affects children with asthma, causing frequent flare-ups, and/or cystic fibrosis (CF) [2]. Invasive pulmonary aspergillosis occurs in immunocompromised children with impaired neutrophil and T-lymphocyte function and children with chronic pulmonary diseases or CF. This severe disease requires prompt treatment, but the diagnosis and management of IPA in children are still challenging and often delayed. Environmental and medical prophylaxis is essential in patients with risk factors, and in such cases, clinical and radiological suspicion is sufficient to initiate empirical therapy with antifungal agents. Since most of the available studies concerning IPA are carried out in adults, we performed a systematic literature review with the aim of summarising the currently available data regarding IPA in pediatric age with a focus on diagnostic tools, treatment and prophylaxis.

The research was restricted to English language. Articles reporting risk factors, diagnostic tools, treatment and prophylaxis for IPA in paediatric population (age < 18 years) with a sample size greater than ten were included independently from the study design. Review articles, case reports, commentaries, editorials, letters to the author and pre-print records were excluded as well as studies referring to adult populations.

Duplicate publications were removed, then two authors separately (RP and MAM) checked the titles and abstracts and removed irrelevant studies according to the inclusion and exclusion criteria. Pertinent articles from the bibliographic references of the selected studies were also considered, and an additional review of the literature was performed prior to final drafting. Articles were categorized according to the study design. Data about IPA risk factors, diagnostic tools, treatment, and prophylaxis were extracted.

The Jadad scale was used to assess quality for RCTs [4], while the Methodological Index for Non-randomized Studies (MINORS) was used for non-randomized ones [5]. Observational studies were evaluated for adherence to Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations [6].

In immune-competent hosts, occasional colonisation of fungal conidia is controlled by the immune system. In contrast, in immune-compromised subjects, an invasive infection can occur, primarily affecting the lungs. However, dissemination to the central nervous system is reported in up to 30% of cases. The IPA development depends on the patient's immune status and lung parenchyma characteristics (Table 2).

IPA represents an emerging problem and is one of the leading causes of morbidity and mortality in immune-compromised patients. The incidence of IPA in children receiving chemotherapy is high. It is associated with increased morbidity and death, with highest rates in patients with acute myeloid leukaemia (AML), recurrent leukaemia, and those undergoing hematopoietic stem cell transplantation (HSCT) [10, 21, 26]. A large multicenter study conducted by Zaoutis et al. in 2006 including 666 children with invasive aspergillosis found that 60–75% of cases are oncologic patients with mortality rates as high as 85% [35]. The risk was higher in case of allogeneic transplantation than in autologous transplantation and in cases of severe graft versus host disease (GVHD). One of the larger cohorts of patients with IPA was described by Burgos et al. in a multicenter retrospective analysis in 2008; they examined 139 paediatric patients with invasive aspergillosis, 80% of which had lung involvement [10]. The most common underlying conditions were haematological malignancies (87/139, 62.6%), followed by inherited immunodeficiencies (16/139, 11.5%), solid organ transplant (16/139, 11.5%), solid tumour (9/139, 6.5%). In particular, 51 patients (26.6%) underwent allogenic HSCT, which was identified as the most critical risk factor for overall mortality in invasive aspergillosis (OR 6.14 – 95% CI 2.67–16.21). Severe neutropenia, defined as neutrophil count below 500 cells/mm³, is the main single risk factor for the development of IPA; in the abovementioned study, it accounted for about 59% of aspergillosis cases. In addition, most hemato-oncology patients presented additional immunosuppression due to prolonged therapy with high-dose steroids or immunosuppressive drugs such as cyclosporine or tacrolimus [10, 26].

Considering primary immunodeficiencies, invasive aspergillosis may be the presenting manifestation or a frequent complication in the first two decades of life in children with chronic granulomatous disease (CDG), a condition caused by neutrophil dysfunction. In such patients, Aspergillus spp. can cause brain abscesses, osteomyelitis and lung damage. Invasive aspergillosis may also occur in children with Wiskott-Aldrich syndrome (WAS), in which neutrophil chemotaxis and lymphocyte function is impaired. Burgos et al. found that CDG accounted for almost 50% of invasive aspergillosis cases among children with primary immunodeficiency (7/16). In contrast WAS was found to be the predominant underlying condition (81/122, 66%) in the larger study by Zaoutis et al., followed by CDG (21/122, 17%) [10, 35]. Among all children with invasive aspergillosis, the highest incidence was observed in those with WAS (30%), followed by CGD (6.5%), allogenic HSCT (4.5%) and AML (3.7%) [35]. Nevertheless, immunodeficiency is not the only risk factor for IPA, since patients hospitalised in intensive care units, children with chronic obstructive pulmonary disease, emphysema, or those affected by chronic diseases such as CF, are also at increased risk. Nevertheless, Zaoutis et al. observed that only 5 out of over 11,000 patients with CF were diagnosed with IPA [35].

The diagnosis of IPA still represents a challenge for the clinician today. Despite the introduction of new diagnostic techniques, the high mortality rate of this condition is mainly due to diagnostic delay [12]. An early diagnosis and prompt therapy is crucial for a better outcome, especially in the immune-compromised child. Maintaining a high index of suspicion in patients with multiple risk factors is essential. An in-depth diagnostic investigation is necessary in case of fever unresponsive to antibiotic therapy, or cough with sputum and dyspnoea in high-risk children. Chest pain with pleural involvement (due to small pulmonary infarcts) and haemoptysis may be associated with the clinical picture of IPA. When infection disseminates to the central nervous system, seizures or radiological alterations consistent with cerebral infarcts, intracranial haemorrhages, or epidural abscesses may occur.

No specific biochemical and/or instrumental tests allow a diagnosis of certainty. Therefore, diagnostic tests should be performed sequentially starting with the least invasive ones. The result of each test should be considered according to the clinical and immune status of the patient. Currently validated diagnostic criteria for adults, which are also used in some paediatric studies [12, 13] distinguish between “proven”, “probable” and “possible” IPA. In a recent Consensus of the European Organization for Research and Treatment of Cancer and the Mycoses Study Group (EORTC/MSG), the definition of invasive fungal disease has been revised, with no change from the previous 2008 classification (Table 3) [36]. These criteria have shown to have an excellent sensitivity but low specificity in detecting possible or probable cases of IPA [7]. The least invasive test for identification of aspergillus infection is the blood assay of galactomannan antigen (GM), a wall component released in the blood due to its growth. GM can also be assayed in other biological fluids, such as bronchoalveolar lavage fluid (BAL) or cerebrospinal fluid, in rare cases of neurological involvement, being more reliable than the blood value [12, 13]. The sensitivity and specificity of the blood assay depend on various components, such as the underlying pathology, current therapies or cut-off used by the laboratory, since there is currently no consensus about negative values. Choi et al. analysed 749 blood samples from 99 oncological children and showed that blood assay of GM was more reliable in cancer patients in detecting invasive aspergillosis (sensitivity: 91.3%; specificity: 81.7%; false positives: 18%) [12]. Similar results were found in the studies of Badiee et al. and Fisher et al. on 62 and 198 paediatric haemato-oncology patients, respectively [9, 17]. Recent studies showed variable sensitivity, specificity and positive predictive value (PPV) of single serum GM determination. On the other hand, consecutive positive tests had higher PPV, especially in a compatible clinical and radiological context [8, 11, 20, 24]. This variability is, to some extent, linked to that of the cut-off used to define GM positivity. In the prospective study of Dinand et al. on 145 neutropenic patients, the optimal cut-off value for single GM determination was found to be 0.7 with sensitivity and specificity both around 82% and negative predictive value (NPV) of 94%, and specificity increased to 91% in case of positivity on a consecutive test [14]. In a recent cross-sectional study of Çağlar et al. on 70 patients with haematological malignancies, consecutive GM positivity displayed higher PPVs independently from the cut-off value chosen [11]. Considering GM on BAL, Mohammadi et al. reported sensitivity and positive predictive values of 87.5% and 93.33% respectively, using a cut-off of ≥ 0.5. Moreover in 7 out of 16 cases of IPA, serum GM was negative, while their BAL GM was positive [32].

Another Aspergillus antigen is β-D-glucan, which is shared with other fungal species such as Candida spp. and Pneumocystis spp. Data available in children are scarce, and an optimal cut-off is unknown, as mean β-D-glucan levels are higher in immunocompetent children than in adults. The combined assay of the two wall components may the reliability of the tests [9] even if, to date, β-D-glucan is not recommended for screening or evaluating suspected IPA in high-risk patients. At last, neither GM nor β-D-glucan (alone or in combination) can be used as a screening marker in neutropenic patients undergoing antifungal prophylaxis, since none of them has shown an acceptable sensitivity [19].

Radiological examinations are essential for the diagnosis of IPA. Nevertheless, recent studies showed that the current radiological criteria used in adults are not applicable in children [10, 12]: radiographic findings considered typical of IPA in adults are not seen in the majority of children with IPA, and unspecific findings are detected more often in immunocompromised children. I In case of multiple nonspecific nodules t chest-xrays (up to 35% of cases) [10], a differential diagnosis with viral (from Cytomegalovirus or Adenovirus) or bacterial (Nocardia) pneumonia is required. Similarly, high-resolution chest computed tomography (HRCT), the most useful imaging tool, rarely evidences pathognomonic lesions of IPA in children, such as “halo sign” (area of ground-glass opacity surrounding a nodule), “air crescent sign” (area of increasing radiolucency in a region of nodular opacity), or cavitary lesions [10, 12]. A single-centre case–control study including 141 children with neutropenic fever lasting more than 96 h showed that the combination use of chest CT scan and serum GM testing was useful for early diagnosis, with PPV and NPV of 76% and 87%, respectively [33]. In the multicenter analysis of Burgos et al., including 110 children with IPA, 61 of them had 2 or more radiological findings on chest CT or plain radiograph, with pulmonary nodules being the most common; (65/110, 59%); on the other hand, halo sign and air crescent were found in smaller percentages of patients (10.9% and 2.2% respectively) [10]. Thus, radiological findings are often unsatisfactory and further examination is required for diagnostic confirmation. Molecular testing through Polymerase Chain Reaction (PCR) on blood or BAL has recently been included among diagnostic criteria for probable aspergillosis. However, its role in patient management has not been established yet. Badiee et al. reported a high NPV of Aspergillus-specific PCR as a screening tool [9], in line with two other studies showing that a combination of GM and PCR testing could be used for screening or diagnostic purposes, especially when tested on consecutive samples [28, 34], However, despite promising results, Aspergillus PCR is not recommended for routine use yet, since its standardization and validation are lacking.

The current gold standard for the diagnosis of IPA is lung biopsy. It is an invasive exam difficult to perform in clinically compromised patients, considering the frequent association between thrombocytopenia and advanced forms of IPA. Therefore, lung biopsy should be performed only when less invasive procedures have not been conclusive. In most cases, lung biopsy is performed through a CT-guided transbronchial route with BAL collection. Since the collected sample is often quantitatively insufficient and the sensitivity of culture examinations is poor, PCR techniques have been introduced to identify Aspergillus DNA on histological samples [9].

The mortality rate of IPA in untreated paediatric patients is close to 100% and remains very high even with aggressive drug therapy. For this reason, empirically based antifungal therapy must be started as soon as possible in children at high risk of developing invasive forms of aspergillosis or otherwise undergoing diagnostic evaluation [7, 27]. In paediatric age, the major difficulties in the therapeutic choice are related to the lack of RCTs on large sample sizes and the paucity of unambiguous data regarding the first-choice drug class, the duration of therapy, and the outcomes to evaluate the clinical response. Most of the available data refer to studies carried out on adults. To date, the first-choice treatment options in children older than 2 years of age are voriconazole, and liposomal amphotericin B. Voriconazole is not approved in children younger than 2 years of age, and the optimal dose is unclear, hence liposomal amphotericin B is the drug of choice. Nevertheless, limited safety data for the use of liposomal amphotericin B in neonates are available. Safety and efficacy of voriconazole were evaluated on 31 paediatric patients treated for 6–9 weeks for invasive aspergillosis in a prospective study showing that voriconazole is generally effective in paediatric patients, with a favourable risk–benefit balance and an overall safety profile similar to adults [31]. However, there is no firm data on the dosing of voriconazole in children aged 2 to 12 years.

In a large multicenter randomized trial on 277 patients older than 12 years old comparing children treated with voriconazole to the ones treated with amphotericin B, the first group was found to have a higher response rate, better survival at 12 weeks of treatment and fewer adverse events [22]. Nevertheless, there is no clear evidence to prefer one of the two drugs, since there is a lack of controlled and randomized comparison studies. On the other hand, there is an unequivocal need to monitor voriconazole plasma levels during therapy, especially in younger patients requiring higher doses [25]; plasma concentrations of 1—5 mg/l are usually considered adequate for prophylaxis and treatment of IPA. The dosages of most common antifungal drugs used for paediatric invasive aspergillosis are summarized in Table 4. Lastly, a few studies have investigated the safety and efficacy of drug combination in children; in a small retrospective study in leukemic patients with invasive fungal disease, 9 of which with invasive aspergillosis, the combination of voriconazole and caspofungin was safe and effective but further studies are needed [27]. Data providing evidence for biologic drug use in ABPA are scarce at the moment [37].

The duration of antifungal therapy has not been established and often needs to be individualized according to the child immune status. In a paediatric population examined in 2007, the mean duration of therapy was 93 days (range 1–880) with a partial response in 45% of cases. However, cases of severely immune-compromised patients in whom therapy was continued indefinitely have also been described [38]. In the prospective study of Martin et al., patients were treated for 6–9 weeks according to clinical response [31]. The most relevant issues in paediatric patients are the sequence of drugs to be used and the clinical parameters to be followed to evaluate the efficacy of therapy. Two main RCTs have been performed in the past in paediatric age groups [29, 39]. Although the expected outcomes are different, they agree that clinical improvement and fever relief for at least three consecutive days are the best evidence of therapy success. Even if regular dosing of GM antigen, after a first positive determination, could be helpful to guide antifungal therapy duration, a negative finding is not sufficient to discontinue therapy [13].

Most of the evidence available in adults suggests the use of posaconazole for prophylactic purpose. It can be used in patients aged older than 13 years, neutropenic, with de novo or recurrent forms of AML, recurrent forms of acute lymphoblastic leukemia, stem cell transplant recipients or affected by GVHD [16] oral voriconazole or itraconazole are the drugs of choice for children aged 2 to 12 years [16, 23, 30]. In patients younger than 2 years of age or unable to take oral drugs, liposomal amphotericin B may be used [15], or, in those older than 2 years, intravenous voriconazole may be prescribed [39]. In a multicenter RCT including 517 patients from 3 months to 30 years of age, caspofungin was found to be superior to fluconazole in reducing invasive fungal diseases, including invasive aspergillosis [18].