FormalPara Take-home message

Invasive pulmonary aspergillosis is an emerging co-infection in patients with influenza who are admitted to the ICU. An international team of experts proposed consensus case definitions of influenza-associated pulmonary aspergillosis in order to facilitate clinical studies and the definition may also be useful to study COVID-19-associated pulmonary aspergillosis.

Introduction

Invasive pulmonary aspergillosis (IPA) is a well-recognized disease affecting immunocompromised individuals with prolonged neutropenia, inherited neutrophil disorders or T cell defects, with the risk depending on the patients’ underlying disease and the type and duration of immunosuppressive therapy [1]. Patients at the highest risk of invasive aspergillosis (IA) include those undergoing intensive chemotherapy for acute leukemia (AL) or recipients of allogeneic cell transplantation (alloHCT) who develop severe graft-versus-host disease, for whom antifungal prophylaxis is currently recommended [2, 3]. With changing treatment modalities, new risk groups continue to emerge, such as patients treated with ibrutinib [4, 5].

Although over the past four decades a link between influenza and IPA has been noted in single cases [6], recent cohort studies provide new insights into the epidemiology and clinical presentation of IPA in intensive care unit (ICU) patients with influenza [7,8,9]. Patients presenting with influenza-associated pulmonary aspergillosis (IAPA) may have classic European Organization for Research and Treatment of Cancer (EORTC)/Mycosis Study Group Education and Research Consortium (MSGERC)-defined host factors [10], but a notable proportion of patients was deemed to be at low risk of IPA, including previously healthy individuals. In addition, the clinical and radiological presentation was often atypical with radiological features that were not considered suggestive of invasive fungal disease. As a consequence, we cannot classify these patients according to existing consensus definitions, i.e., the EORTC/MSGERC definitions and the AspICU algorithm for classification of IPA patients in the ICU [10, 11]. We therefore set out to discuss current insights into the epidemiology, pathogenesis, diagnosis and management of IAPA and to propose case definitions that can facilitate homogeneity and comparability in clinical studies.

Participants and methods

The expert panel is comprised of 29 participants from seven European countries, the USA and Taiwan. To ensure heterogenicity, participants were selected from various fields of expertise: medical microbiology (PEV, KL, CL-F, TRR), infectious diseases (BJAR, MB, TC, CJC, OAC, DRG, NAFJ, BJK, OL, MH-N, TFP, FLvdV), intensive care medicine (EA, SB, PD, PW-LL, IM-L, JAS, LV, JW), clinical pharmacology (RJMB, RL, IS), public health (TC) and hematology (OAC, JM). Selected participants furthermore had specific expertise in epidemiology, diagnosis and management of invasive fungal diseases or fungal disease guideline development. The meeting was prepared by PEV, RJMB, JW and FLvdV. Case definitions were developed through a process of informal consensus. Although a systematic literature review was not performed, experts in the field presented overviews regarding epidemiology, pathogenesis, diagnosis and treatment of IAPA, which were followed by a group discussion process designed to allow members of the group to voice their opinions and contribute equally to the decision-making [12]. The goal of the consensus process was to bring the group to general agreement. Presentations and initial discussions took place on a single day meeting in April 2019 in Amsterdam and were continued through electronic exchange of views until consensus was achieved. The chosen framework included host and risk factors, clinical factors and mycological evidence, similar to the framework in the EORTC/MSGERC definitions and the AspICU algorithm [10, 11]. A medical writer made notes of the meeting, which were used as input to write the manuscript. A first draft manuscript was prepared by PEV, BJAR, RJMB, JW and FLvdV and circulated for comments from all experts. The experts reviewed and commented on the manuscript. Using these comments, a final version was circulated for approval. The logistics of the meeting were handled by a certified Congress organizer (Congress Care, s’Hertogenbosch, the Netherlands) with financial support of Pfizer (Pfizer B.V., Capelle aan den IJssel, the Netherlands). Congress Care and Pfizer had no influence on the selection of participants, selected topics, discussions, preparation and final approval of the content of the manuscript.

Expert review

Global epidemiology of influenza and IAPA

Although figures vary depending on geographic region, season and vaccination rates, approximately 0.1% of influenza patients require hospital admission with 5–10% of these requiring ICU admission [13, 14]. The mortality in patients admitted for influenza is 4% and 20–25% for those admitted to ICU [14,15,16]. Bacterial superinfection is common, affecting 10–35% of cases, typically with Streptococcus pneumoniae or Staphylococcus aureus [16]. However, a recent Dutch–Belgian multicenter study over seven influenza seasons in seven institutes demonstrated influenza as an independent risk factor of IPA (adjusted odds ratio 5.19, 95% confidence interval (CI) 2.63–10.26, p < 0.001) [9]. Results also showed that the 90-day mortality rate for ICU patients with IAPA was almost double that of ICU influenza patients without IAPA (51% vs. 28%, adjusted odds ratio 1.87, 95% CI 1.05–3.32). IAPA was initially thought to be associated with influenza A/H1N1pdm09 only [7, 17], but it became clear that IAPA is also associated with other influenza A and influenza B viruses [9]. The median time between influenza diagnosis and IAPA was short, often in the first 5 days [7, 8, 18]. Studies have shown considerable variation in rates of IAPA in different countries, with high rates in the Netherlands, Belgium and Taiwan, but lower rates in other countries [19], and in some we do not know the incidence (e.g., USA) [20,21,22]. Potential reasons for these regional differences are related to the underlying conditions, concomitant exposure to corticosteroids, environmental factors, including exposure to Aspergillus, use of non-culture-based diagnostic tests for Aspergillus (e.g., galactomannan (GM)) and differences in awareness of IAPA [23,24,25]. Autopsy rates are very low, which results in a considerable underdiagnosis in many countries [26]. Other factors that might contribute to regional differences in IAPA rates include influenza vaccination rates, with different policies in different countries, and differences in influenza antiviral treatment strategies with oseltamivir or zanamivir [27]. Annual vaccination reduces influenza-associated complications (hospitalization, ICU admission, severity of illness, superinfection) and improves the outcome in transplant recipients and COPD patients [28, 29].

Pathogenesis of IAPA

In pending studies that explore the pathogenesis of IAPA and host immune defects, it is likely that damage to the epithelium by influenza and defective fungal host responses in the lung due to influenza and/or inflammatory conditions predispose to Aspergillus disease, similar as what is seen in bacterial superinfections. Furthermore, autopsy studies have shown the presence of sporulating heads of Aspergillus inside the bronchi with invasive growth occurring into the lung tissue. Sporulation could contribute to a high fungal burden and spread of the disease within the lung, thus contributing to the rapid disease progression and extensive lung damage. Other factors that have been implicated in IAPA include the use of corticosteroids and of neuraminidase inhibitors, such as oseltamivir [7, 30]. Ultimately, these insights may aid in identifying patients at risk of IAPA and to design effective antifungal and adjunctive immunomodulatory treatment strategies.

Clinical presentation and diagnosis of IAPA

A retrospective Belgian study of influenza patients admitted to ICU between September 2009 and March 2011 showed that 9 of 40 (23%) patients had IAPA. Four cases (44%) were proven despite not being immunocompromised according to the EORTC/MSGERC consensus definitions [7]. The median time between influenza diagnosis and IAPA was 2 days (range 0–4 days). All IAPA patients had positive BAL GM, and 78% had positive serum GM, despite not being neutropenic. Eighty-nine percent of patients had Aspergillus growth in BAL culture (almost exclusively Aspergillus fumigatus), and 55% of patients had endobronchial lesions observed during bronchoscopy, possibly indicating invasive tracheobronchitis [7]. Similar performance characteristics of BAL GM and culture were reported in two other cohort studies [8, 9]. BAL sampling is thus an important diagnostic procedure as serum GM can be negative and sputum/tracheal aspirate cultures can remain sterile.

Lesions that are suggestive of invasive mold disease on imaging in neutropenic patients, such as the halo sign, are often absent in critically ill patients. However, in some IAPA patients with autopsy-confirmed Aspergillus tracheobronchitis, chest CT demonstrated peribronchial infiltrates. The main diagnostic clue for airway-invasive Aspergillus tracheobronchitis is epithelial plaques, pseudomembranes or ulcers that can be visualized via bronchoscopy, as radiological features may be subtle [31]. Worsening of radiographic pulmonary infiltrates in patients with influenza is often attributed to progression of ARDS or bacterial infection, leading to a change of antimicrobial therapy without performing diagnostic procedures [32]. Patients who survived IAPA received antifungal therapy much earlier than those who did not (2 days after diagnosis of influenza among survivors versus 9 days among non-survivors) [8], suggesting that early diagnosis and administration of antifungal therapy may be important. Lateral flow tests have recently become available as an alternative for diagnosing IPA (AspLFD, OLM Diagnostics and the sōna Aspergillus GM, IMMY) showing overall good performance in hematology patients [33]. The very quick assessment, with results available within 30–45 min, makes this type of test very attractive for the management of IAPA and use in clinical trials. However, lateral flow tests have not yet been validated in the ICU population.

IAPA needs to be considered in patients admitted to the ICU with influenza and where indicated these patients should undergo early BAL for Aspergillus antigen testing, culture and microscopy. Patients who test positive require anti-Aspergillus therapy, and the BAL fluid sample should be fast-tracked for azole resistance testing by PCR (and culture when positive) in regions with high (> 5%) azole resistance rates [34]. This would enable diagnostic assessment and initiation of adequate antifungal therapy within 24–48 h of ICU admission. Diagnostic workup for IAPA may be repeated in patients deteriorating while on antivirals and/or appropriate antibiotics or when initiating corticosteroid treatment is unavoidable.

Discussion on clinical presentation and diagnosis of IAPA

If a patient is admitted to the ICU and has influenza with pulmonary infiltrates, the diagnosis of IAPA should be considered and further investigation performed as appropriate. Ideally, this would include in order of invasiveness, serum GM testing, fungal cultures of sputum and/or tracheal aspirate, pulmonary CT, bronchoscopy to visualize the large airways and obtain BAL fluid for GM testing and fungal and bacterial cultures. Testing is most appropriate in patients who are on mechanical ventilation, but the diagnostic strategy is less clear in patients not intubated. As up to 50% of patients may present with tracheobronchitis, the presence of plaques and ulceration might be considered for inclusion in the definition of IPA [35]. Policies for taking biopsies of lesions seen on bronchoscopy may vary, mainly because of concerns about the risk of bleeding with biopsy in ICU patients. The use of a flexible brush may also be sufficient to make the diagnosis.

Although a positive serum GM is highly indicative of IA, BAL GM can be positive in patients with Aspergillus colonization. It therefore does not absolutely discriminate between colonization and invasive disease. However, it clearly makes it more likely that an invasive disease is present [36].

Use of corticosteroids

Corticosteroids should not be given to influenza patients as their use may be associated with increased risk of IAPA [7, 37,38,39]. A recent Cochrane review on this topic concluded that the use of corticosteroids in patients with influenza was associated with a worse outcome [40]. However, the evidence was almost exclusively observational. Furthermore, patients are often given steroids in the first few days preceding or after ICU admission for a variety of reasons including COPD exacerbation or complications such as sepsis. With surveys suggesting that approximately half of the physicians are not aware of IAPA [24], many physicians may additionally not be aware of the potential drawbacks of corticosteroids. Whenever the use of corticosteroids is unavoidable, more efforts (bronchoscopy with GM detection in BAL fluid or serum β-D-glucan test) should be made to exclude or diagnose IAPA [41].

Rationale for antifungal prophylaxis for IAPA

In settings with high IAPA rates in ICU patients with influenza pneumonia, an antifungal prophylaxis strategy might be appropriate, particularly as IAPA typically occurs early after ICU admission. However, there is currently no mold-active antifungal agent licensed for prophylaxis of IA in ICU patients. Posaconazole (POS) prophylaxis reduces the prevalence of IA in neutropenic AML patients and those with graft-versus-host disease following alloHCT [2, 3]. Based on this proof-of-principle, it has been hypothesized that POS prophylaxis can reduce IAPA prevalence in ICU patients. Intravenous (IV) administration of POS prophylaxis in the ICU is favored in patients on mechanical ventilation or with a high likelihood of malabsorption of oral formulations. POS IV formulation should be administered through a central catheter due to its acidity (pH 3.2) [42].

Treatment options and challenges for IAPA in the ICU

First-line treatment options for IPA include voriconazole and isavuconazole [35, 43]. Other options include echinocandins in combination with anti-mold azoles, and liposomal amphotericin B (L-AmB) in regions with high rates of azole-resistant A. fumigatus, although clinical data with L-AmB in ICU patients are limited [43, 44]. Achieving adequate drug exposure is challenging in ICU patients with multiple factors contributing to pharmacokinetic variability. Unlike L-AmB and the echinocandins, drug interactions are clinically relevant for the azoles and pharmacogenetic factors are important in inter-individual drug exposure variability [45]. The impact of therapeutic drug monitoring (TDM) for voriconazole shows a clear relation between exposure and both efficacy and toxicity. Target plasma trough voriconazole concentrations of ≥ 1.5–2 mg/L are associated with near-maximal clinical response in treatment of IA with a wild-type phenotype [46,47,48,49,50,51], with higher exposures (> 5.5 mg/L) increasing the risk of (neuro)toxicity. Higher trough concentrations (> 2 mg/L) are recommended for treatment of pathogens with elevated MICs (e.g., > 0.25 mg/L) [52]. For isavuconazole, there is no robust target plasma concentration, and the population average exposure of participants that demonstrated a favorable response (2–4 mg/L) is commonly used [43].

Discussion on antifungal treatment options and challenges for IAPA in the ICU

A specific drug–drug interaction is relevant for patients with IAPA given the fact that co-infections with S. aureus are frequently observed; undetectable voriconazole levels have been observed in 11 of 20 patients, who were concomitantly treated with flucloxacillin [53], but the mechanisms of interaction are not yet fully understood. Similar interactions have not been seen with other azoles. Many other drug interactions with azoles and drugs commonly deployed in the ICU can be expected [45].

Aerosolized antifungal treatment may be a useful adjunctive therapy to systemic antifungal therapy for patients with confirmed Aspergillus tracheobronchitis, to achieve good endobronchial exposure [35, 54]. However, dense lipophilic plaques in the trachea may be difficult to penetrate and more research is needed into when and how to use aerosolized antifungals as well as their efficacy. The ECCMID/ECMM/ERS Aspergillus guideline reviewed the teratogenic and mutagenic potential of antifungals in early pregnancy and recommends that azoles should be avoided, with polyenes being considered the preferred therapy [43, 55]. Thus, for pregnant patients at risk of IAPA a diagnostic approach was preferred above antifungal prophylaxis.

There is little evidence on the impact of ECMO on antifungal drug exposure [56]. For the echinocandins, an impact of ECMO is not expected. Experts felt that, given these uncertainties, TDM of any antifungal used would be advised to ensure sufficient drug exposure.

Consensus case definition for IAPA

The expert panel discussed which case definition of IAPA would be appropriate to use in clinical studies, initially considering various aspects regarding four main areas of focus: entry criteria of the consensus definition, host, clinical features and mycological evidence similar to the currently used EORTC/MSGERC classification.

Entry criterion

In addition to having a positive diagnostic test for influenza, patients would require to have a clinical syndrome compatible with influenza disease as part of the definition. This criterion should be termed the ‘entry criterion’ and not ‘host factor’ for clarity. To avoid the risk of missing patients who initially tested negative with a rapid influenza antigen test but subsequently tested positive (by PCR) for influenza when admitted to hospital, a recommendation on a timescale, such as between 1 week before ICU admission and 72–96 h post-admission, was included. The consensus on the entry criterion was: a patient requiring ICU admission for respiratory distress with a positive influenza PCR or antigen test temporally related to ICU admission.

Host factors

Host factors are considered in the EORTC/MSGERC definition and AspICU algorithm [10, 11], but the system of taking host factors into account was a necessity because the risk of a false-positive Aspergillus test increases substantially when the test is done in patients at low risk of the disease. Clinicians had to take into account the type of host in order to increase the pretest probability of an invasive fungal disease being present. However, for IAPA the key question is whether the disease is present or not, and not whether the patient group has a higher risk than other patient groups for developing the disease. More importantly, the incidence of IAPA in patients admitted to the ICU with influenza may be higher in some centers [9, 21]. No further host factors are needed to increase the pretest probability in this patient population. Although most IAPA cases have at least one underlying condition or steroid use, host factors were not be included in the case definition for IAPA.

Criteria to define proven and probable cases of IAPA

The distinction between proven and probable IAPA is important for clinical trials, while in clinical practice, people should not distinguish between proven and probable disease.

The criteria for proven disease include a patient fulfilling the entry criterion plus histological evidence of invasive fungal elements and mycological evidence for the presence of Aspergillus (obtained by Aspergillus PCR or culture from tissue). Tracheobronchitis (tracheal and/or bronchial ulcerations or nodules, pseudomembranes or plaques visualized at bronchoscopy), as also described in the EORTC/MSGERC definitions [10], is a separate entity. Although a tissue biopsy would normally be required to prove a case of IAPA, in tracheobronchitis cases hyphal elements suggestive of Aspergillus seen on sloughed-off pseudomembrane, and Aspergillus identified on culture or PCR, can also be considered proven disease (Table 1).

Table 1 Proposed case definition for IAPA in ICU patients
Full size table

A patient fulfilling the case definition of probable IAPA is required to fulfill the entry criterion. A positive serum GM (GM index > 0.5) is important evidence for the diagnosis of IAPA, in patients with pulmonary infiltrates on chest X-ray or other imaging modality or bronchoscopic evidence of tracheobronchitis (Table 1). In patients with tracheobronchitis, an infiltrate is not required.

In patients with endobronchial plaques or pulmonary infiltrates, a positive BAL GM or culture of a tracheal aspirate is considered mycological evidence that supports a probable IAPA diagnosis. In patients with bacterial pneumonia where Aspergillus is cultured only from a sputum sample, there may be a risk of overdiagnosis and thus over-treatment. For clinical practice, clinicians should take into account that a positive culture of an upper airway sample may indicate IAPA, but that confirmation with serum or BAL GM or BAL culture should be pursued. However, one problem is that the background incidence varies in different regions, making it difficult to develop generalized guidelines that apply uniformly. The significance of a positive sputum culture thus depends on the background incidence in a specific unit. Although any Aspergillus-positive respiratory sample is in itself insufficient to classify patients as probable IAPA, a new pulmonary cavitating infiltrate is indicative of IAPA in patients who meet the entry criterion. Therefore, any Aspergillus-positive respiratory sample is sufficient evidence to classify patients as probable IAPA provided that a pulmonary cavitating infiltrate is present (Table 1; Fig. 1).

Fig. 1
figure 1

Flowchart of probable IAPA classification. (*)If hyphae consistent with Aspergillus are documented in a biopsy of an airway lesion AND Aspergillus is grown from sputum or a tracheal aspirate, the case fulfills the definition of proven IAPA

Full size image

A BAL GM index cutoff of ≥ 1.0 is recommended as this cutoff value ensures high specificity, without decreasing sensitivity significantly, which is also in line with other definitions and recommendations [10, 57]. Aspergillus PCR is not recommended as a primary diagnostic tool because of concerns about its reliability and positive predictive value for the diagnosis of IPA. However, Aspergillus PCR is recommended in the proven category because it enables Aspergillus identification in tissue samples.

In some patients, discordant results are obtained, for instance a positive sputum culture but negative BAL GM. For most situations, IAPA classification relies on a positive GM test, as a positive sputum culture with a negative GM result would be interpreted as a lower probability of IAPA (unless a pulmonary cavity or tracheobronchitis is present)(Fig. 1).

Conclusion

IAPA has emerged as a severe complication of influenza, especially in ICU patients, and this secondary infection may occur in any patient, including those considered to be at low risk of developing IPA. The global epidemiology of IAPA may be variable, which might be partly due to underdiagnosis [24]. The clinical presentation of IAPA includes invasive Aspergillus tracheobronchitis, which requires bronchoscopic visualization of plaques in the airways to make a diagnosis. Aspergillus culture and BAL GM are positive in > 80% of IAPA cases, and ordering such tests is recommended in influenza cases in the ICU. The proposed case definition relies on an entry criterion based on an influenza-like illness and the detection of influenza virus. The case definition distinguishes between invasive tracheobronchitis and other pulmonary forms of IAPA, with demonstration of invasive fungal hyphae with positive mycology qualifying as proven infection. Detection of GM or positive Aspergillus culture in BAL is the main mycological criteria in probable case definition.

The expert group acknowledges that to date still limited data exist to support a definitive approach regarding definitions, diagnosis and treatment of IAPA, but the proposed case definition will facilitate clinical research, will enable valid study comparisons and is essential for surveillance. Awareness of IAPA and early antifungal therapy based on high clinical suspicion and Aspergillus diagnostics remains critical to improve the outcome of IAPA.

Can the IAPA definitions be applied to COVID-19-associated pulmonary aspergillosis?

Recent reports of IPA cases in coronavirus disease 2019 (COVID-19) patients in the ICU raise the question of whether these IAPA definitions can be applied to COVID-19-associated pulmonary aspergillosis (CAPA) [58,59,60]. Although the number of CAPA cases that have been reported is still limited, two recent studies reported putative CAPA cases in 9 of 27 (33%) and 5 of 19 (26%) COVID-19 patients admitted to the ICU [59, 60]. Although the high number of cases suggests a high risk of developing IPA in COVID-19 patients, there are a number of differences regarding the pathogenesis of SARS-CoV-2 infection compared with influenza (Table 2). In influenza patients, there are several factors that are thought to contribute to the risk of IAPA, including the local tissue damage caused by influenza, an immune modulatory effect by suppression of the NADPH oxidase complex and possible effect of treatment with neuraminidase inhibitors, such as oseltamivir. In SARS-CoV-2 infection, another receptor is used by the virus to enter human cells, which are not commonly found in the large airways (Table 2). Thus, the risk of invasive Aspergillus tracheobronchitis may be lower in CAPA compared with IAPA. In addition, there is no known direct immune modulatory effect of SARS-CoV-2, which suggests no virus infection-related increased risk of IPA. While IAPA is characterized by rapidly fatal infections with high fungal burden, such course of disease progression has not been reported for CAPA. On the contrary, eight of nine CAPA cases reported from a French cohort did not receive antifungal therapy, with a mortality rate similar to COVID-19 cases without IPA [59]. As, in contrast to IAPA cases, virtually all CAPA cases reported to date are serum GM negative, the question remains if COVID-19 patients develop invasive disease or just become colonized with Aspergillus. It is possible that COVID-19 is in itself not a risk factor for IPA, but that the risk is associated with other risk factors related to treatment such as administration of corticosteroids or underlying host factors. Nevertheless, the high rate of Aspergillus recovered from COVID-19 patients suggests that there might be conditions that favor growth of the fungus in the lung. We think that the proposed IAPA case definitions may be considered for classification of CAPA patients, while awaiting further histopathological studies that provide more insight into the interaction between Aspergillus and the SARS-CoV-2-infected lung.

Table 2 Comparison between characteristics of IAPA and CAPA
Full size table