The diagnosis of invasive pulmonary aspergillosis (IPA) in intensive care unit (ICU) patients is crucial since most clinical signs are not specific to invasive fungal infections. To detect an IPA, different criteria should be considered. Next to host factors and radiological signs, microbiological criteria should be fulfilled. For microbiological diagnostics, different methods are available. Next to the conventional culture-based approaches like staining and culture, non-culture-based methods can increase sensitivity and improve time-to-result. Besides fungal biomarkers, like galactomannan and (1→3)-β-D-glucan as nonspecific tools, molecular-based methods can also offer detection of resistance determinants. The detection of novel biomarkers or targets is promising.
1. Introduction
Invasive pulmonary aspergillosis (IPA) is most common in patients with hematological malignancies, like acute myeloid and lymphoid leukemia (AML, ALL). But other immunosuppressive diseases or distinct circumstances, like a pulmonary virus infection or a long stay at the intensive care unit (ICU), are risk factors for an IPA as well
[1][2][3][4][5]. In a multicenter study, a multivariate analysis to determine the risk of death was performed. Of 152 patients with invasive fungal infections, 92 (60.5%) died
[6]. The diagnostic of IPA is crucial due to various criteria, which are non-fungal-specific. The infectious diseases group of the European Organization for Research and Treatment of Cancer and the Mycoses Study Group (EORTC/MSG) defined three classes for the IPA (proven, probable and possible), depending on which consensus criteria are fulfilled
[7]. These criteria include host factors of the patients (e.g., previous illnesses or immunosuppressive therapy), clinical features (radiological signs, abnormalities in typical anatomical sides) and mycological evidence (culture or microscopical detection of fungal structures in respiratory material and detection of biomarkers, like galactomannan (GM) antigen or (1→3)-β-D-glucan (BDG) and DNA detection)
[7]. As mentioned by Bassetti, patients in ICUs often do not have risk factors like hemato-oncologic patients. In these patients, risk factors for aspergillosis are treatment with corticosteroids in high doses, diabetes, chronic liver or pulmonary diseases (COPD), or malnutrition. Furthermore, influenza and other viral infections like COVID-19 are risk factors
[8]. Typical radiological signs of aspergillosis are usually not found in ICU patients. For diagnosing invasive fungal diseases (IFD) in critically ill patients, Blot et al. defined criteria to discriminate between colonization with
Aspergillus in the respiratory tract from an IPA. Here, only two categorizations are defined, the proven IPA (same as EORTC/MSG criteria) and the putative IPA
[9]. Additionally, Schauwvlieghe et al. also defined a clinical algorithm to diagnose IPA in critically ill patients and Koehler et al. in patients with COVID-19
[3][4]. Finally, Bassetti et al.
[8] defined a probable invasive aspergillosis in patients at risk (glucocorticoid treatment with prednisone equivalent of 20 mg or more per day, or ≤500 neutrophil cells/mm
3, or COPD, or treatment with immunosuppressants, or hemato-oncological diseases, or organ transplantation, or HIV-infection, or influenza or COVID-19) in combination with the microscopic or cultural detection of
Aspergillus spp. in a sample of the lower respiratory tract and/or a GM index of >0.5 in serum and/or >0.8 in bronchoalveolar lavage (BAL).
Several laboratory tests for diagnosing IPA are commercially available, e.g., fungal stains, culturing, biomarkers and molecular methods
[7]. For any of these test methods, different providers use distinct techniques. These differ regarding their statistical/analytical values to diagnose IPA. In the last couple of years, new biomarkers, as well as the use of new targets, have been described for the detection of invasive aspergillosis (IA) in critically ill or high-risk patients, contributing to a more reliable diagnosis of IPA.
2. Non-Culture-Based Diagnostic Tools to Detect an IPA
2.1. (1→3)-β-d-Glucan
Non-culture-based methods, e.g., GM and BDG, reduce the time needed to identify IPA in comparison to culture-based methods. Whereas GM from both serum and BAL samples is recommended as a reliable marker in the early detection of IPA
[4][7], EORTC/MSG consensus definitions of invasive fungal diseases (IFD) include BDG as mycological evidence while not recommending it for use in clinical trials or for defining IPA as BDG is not specific for any invasive fungal disease
[7]. For other specimens like BAL, data are heterogeneous, and no official recommendations by organizations or consortia exist.
BDG is a major cell wall constituent of most medically important fungal species
[10]. During IFD, it is released into blood and tissues. Exceptions are cryptococci, Zygomycetes (such as
Absidia,
Mucor and
Rhizopus) and
Blastomyces dermatitidis which are known to have little or no BDG and thus, glucan is not detected during infection with these organisms
[10][11][12].
In a meta-analysis from 2016, Shi et al. reported a much lower sensitivity and specificity of BAL BDG measurement than those of serum BDG (52% and 58%, respectively) with the limitation of high heterogeneity
[13]. Study cohorts included patients with and without immunosuppression as well as from ICUs and general wards. Even after the exclusion of the outlier study and reappraisal, performance characteristics were still poor (45% sensitivity and 62% specificity)
[13].
2.2. Galactomannan
There is plenty of data on the use of distinct GM assays in BAL samples for the identification of IPA. Three different test procedures for measurement of GM are currently available. The classic GM assay is an enzyme-linked immunosorbent assay (EIA), a chemiluminescence immunoassay (CLIA) and the lateral flow assay (LFA). The GM assays are available from Platelia (Bio-Rad Laboratories, Marnes-la-Coquette, France) and Euroimmun AG (Lübeck, Germany), the CLIA from Vircell S.L. Parque Tecnológico de la Salud (Granada, Spain) and the LFA from OLM Diagnostics (Newcastle Upon Tyne, UK) and IMMY (Oklahoma, OK, USA).
2.2.1. Testing of Sputum Samples
Among others, the detection of GM in sputum samples has a special indication from patients with cystic fibrosis. In this context, the determination of GM serves along with other test results (e.g., total IgE, anti-
A. fumigatus IgG and IgE, PCR) in a classification of
Aspergillus disease (bronchitis, colonization, sensitization)
[14]. There is only one study in which GM was investigated in the specimen “sputum” of hematological patients. With a cut-off index of 1.2, sensitivity was 100% compared to 66.7% in BAL and 83.3% in serum samples. The specificity was 62.2%
[15].
In patients with COPD, the sputum GM OD index was higher in those with bronchiectasis (index 3.7) than in patients without bronchiectasis (0.7). Sputum GM detection correlated with a positive
Aspergillus culture and PCR results
[16]. In another study, sputum samples of patients with allergic bronchopulmonary aspergillosis (ABPA) (
n = 33) and chronic pulmonary aspergillosis
n = 126) were investigated by using culture, PCR and GM. The culture was positive in 13%. Depending on the cut-off index (1.0, 4.5, 6.5), the sensitivity of the GM test was 87%, 67%, and 69%, specificity was 31%, 65% and 80%
[17].
2.2.2. Testing of BAL Samples
The role of GM testing in BAL has been extensively studied in hemato-oncological patients. In a meta-analysis, a sensitivity of 85% and a specificity of 86% were found when using a cut-off index of 1.0
[18]. The updated EORTC/MSG consensus definition recommends a cut-off index of 1.0 for GM EIA (Platelia) in BAL samples, also with a sensitivity of 75–86% and a specificity of 94–95%. Sensitivity and specificity were similar whether or not haematological malignancies are the underlying disease (85%/87% sensitivity and 91%/89% specificity). If serum/plasma shows an OD ≥ 0.7, a BAL result of ≥0.8 should be interpreted as a relevant result, also
[19].
In influenza-related aspergillosis a proposal for a case definition from 2020 recommended a BAL cut-off of 1.0
[20]. In COVID-19 patients, only restricted use of bronchoscopy has been recommended at the beginning of the pandemic to protect the BAL-sampling staff from the potential risk of infection due to aerosol generation. In this context, other materials were preferred in these patients, for example, tracheal secretion. Also, for this material, a cut-off of ≥1.0 was used
[21]. Furthermore, in ventilated patients, microaspiration of GM-containing enteral nutrition might be a disruptive factor.
The GM-EIA and GM-LFA for BAL samples were analyzed in a study of CAPA
[22]. All studied samples were classified as probable CAPA. GM-LFA in BAL showed similar diagnostic capabilities to the classic GM assay, but it is faster for diagnosing CAPA. GM-LFA showed a sensitivity of 60.6%, specificity of 88.9%, PPV of 71.4% and NPV of 83.1% when compared with BAL culture, respectively. GM-EIA showed a sensitivity of 54.5%, specificity of 91.7%, PPV of 75%, and NPV of 81.5% for BAL samples, respectively. Results show that the GM-LFA can be used as an alternative approach in the absence of GM-EIA testing
[22].
2.3. Aspergillus PCR Assay in BAL
The PCR assay is one of the most commonplace detection methods in clinical settings. Several publications from the past years have discussed the performance of the PCR assay in BAL as a diagnostic tool for IPA in different patient cohorts. In a review of commercially available PCR tests, sensitivities of 68–94% and specificities of 80–98% were found in different patient groups
[23].
Scharmann et al. evaluated three PCR assays for the detection of IPA in immunocompromised patients
[24]. In this study, the variation between different manufacturers becomes clear. Here, statistical analysis revealed a variation in the sensitivity from 60.0% to 80.0%, the specificity from 73.2% to 96.7%, the same as the PPV varies between 26.7% and 70.0% and NPV between 95.4% and 96.8%
[24].
2.4. Combination of Diagnostic Methods
Since the emergence of CAPA and its high mortality rates, the need for an efficient diagnosis of IPA has become evident
[25]. Some studies have published results on the combined diagnostic capabilities of the GM and PCR assays in non-hematological patients, which is significantly improved compared to the use of one of these methods solely.
A study of 63 CAPA patients where BAL samples were analyzed assessed the importance of
Aspergillus species in ventilated patients. Here, assays such as BAL-GM/ serum-GM or BAL-PCR were used. Probable CAPA was diagnosed in 17% of patients, not all of whom had EORTC/MSG host factors for IPA. Sensitivity (range) for PCR, BAL GM and serum GM was 44% (13.7–78.8), 55.6% (21.2–86.3) and 33.3% (7.5–70.1), respectively.
2.5. Novel Biomarker/Targets
The knowledge of the pathogenesis of IPA identifies the interaction of the ciliated epithelium and innate immune system, including resident alveolar macrophages and dendritic cells, and recruited inflammatory cells, as the first line of defense against inhaled fungal spores
[26]. These cells express a large repertoire of immune receptors, sensing pathogen motifs and driving the secretion of cytokines and chemokines that control innate and adaptive immune responses
[27].
The question is if a specific inflammatory signature of cytokines and chemokines is specific for IA. In one study, cytokine profiles in BAL samples from patients with IPA were compared to matched control patients
[26]. It was shown that a subset of alveolar cytokines could significantly discriminate cases of
Aspergillus infection from those without infection. Furthermore, it was reported that two distinct clusters of highly correlated cytokines (IL-1β, IL-6, IL-8, IL-17A, IL-23, and TNFα) were differentially expressed between cases of IPA and controls. IL-8 was the best-performing cytokine, with alveolar levels ≥904 pg/mL predicting IPA with elevated sensitivity (90%), specificity (73%), and NPV (88%). This was the first study of its kind, including BAL samples. High serum IL-8 levels were reported as a reliable blood biomarker for IPA
[28].
Cytokines also play a major role in ICU patients with COVID-19 and are a risk for developing CAPA. The cytokine storm that occurs in COVID-19, instead of activating a competent immune response to possible opportunistic infections, causes the dysregulation of the immune system. This is linked to a reduction of lymphocyte count, dampening of cell-mediated fungicidal activity and ineffective conidial killing, creating a fertile ground for fungal invasion. The cytokine pattern expressed in severe COVID-19 shares some similarities with severe IPA (i.e., high levels of TNF-α, IL-1, IL-6, IL-8, IL-10, and low levels of IFN-γ)
[29].
TAFC can also be found in urine. The urine TAFC, normalized to creatinine, was measured in high-risk patients
[30]. TAFC/creatinine sensitivity, specificity, and positive and negative likelihood ratio for probable versus no IPA (cut-off ≥3) were 0.86, 0.88, 6.86 and 0.16 per patient. This approach shows the advantage of non-invasive sampling. The determination of the siderophore TAFC was performed in most studies by mass spectrometry. Mass spectrometry is not available everywhere and needs expertise and experience.
3. Conclusions
Taking everything into account, it is crucial to detect an IPA reliably in ICU patients. Based on recommendations from EORTC/MSG criteria, Blot et al., Bassetti et al., and Schauwvlieghe et al., not only host factors and clinical signs should be fulfilled, but also microbiological diagnostic is an important milestone for IPA diagnostics
[7][9][31]. Various studies investigated the diagnostic tools with different results, which led to the observation of non-homogeneous patient groups. A comparison of the different studies is critical due to the different cohorts observed. The combination of different methods and the investigation from different specimens (BAL and serum samples) seems to bring the most reliable results. BDG in serum shows a high NPV, while GM EIA from BAL samples showed the highest specificity for ICU patients. The cut-off, which is well established in hemato-oncological patients (index ≥1.0), is not so clear in other patient groups. The cut-off may differ in different patient groups in the future. The GM LFA showed a lower specificity than the EIA, but at the same time, it is faster and more easy to handle. Detecting
Aspergillus with a commercially available PCR, especially the AsperGenius, which has been shown to be the best-evaluated method, showed high specificity and NPV in all patient groups. So far, new biomarkers or targets (e.g., cytokines, TAFC or secondary metabolite) are not clinically investigated sufficiently, making evaluation of these biomarkers necessary in the future.