Symptoms of invasive pulmonary aspergillosis (IA) are not specific and it is therefore recommended to adjust the clinician’s awareness according to the risk level of their patients in order to prescribe prompt and relevant further investigations. Fever refractory to, or recrudescent despite, antibiotics is often the first and sometimes the only symptom of IA. Although frequent, fever can be missed in some patients, for instance in those receiving corticosteroids. Cough, sputum production, hemoptysis, pleuritic chest pain or rub, dyspnea or bronchospasm are associated with pulmonary localization of IA.

Chest radiograph is an obsolete procedure whenever chest computed tomographic (CT-scan) is available ^[1][2][98,99]. Contrast agents are usually not necessary for diagnosis but can be useful in assessing the risk of hemoptysis if a lesion is close to a large vessel. Depending on its mechanism of extension, IA can present with two different CT-scan patterns: angio-invasive pattern and airway-invasive pattern ^[3][100].

The angio-invasive pattern is the radiological translation of the penetration of hyphae through the vessels wall, leading to fungal thrombi causing necrosis and hematogenous dissemination. It has been classically associated with severe neutropenia and consists of nodules often surrounded by ground-glass opacity, translating perilesional hemorrhage and forming of the halo sign ^[4][101]. The halo sign is considered highly suggestive of neutropenia-related IA, but it is not specific and can be seen is various conditions such as other invasive mold diseases (IMD), bacterial infections, primary or metastatic cancers, granulomatous with polyangiitis, bronchiolitis obliterans with organizing pneumonia, or focal lung injury. It is also important to take into account the potential variations of this typical lesion through time: an early CT-scan might show micronodules, whereas a late CT-scan might miss the halo of ground-glass opacity, that classically disappears after 5 to 10 days ^[5][102]. In an even later setting, the recovery from neutropenia induces a necrotic retraction of this primary lesion and gives the aspect of an air crescent in a nodule. This air crescent can eventually widen into a cavity, first thick-walled and finally thin-walled.

The airway-invasive pattern is mostly seen in non-neutropenic patients ^[3][5][100,102]. Its CT-scan reflects the endobronchial dissemination of infection with thickened bronchial walls with multiple centrilobular nodules, resulting in the typical tree-in-bud pattern. This pattern is widely shared by other pulmonary infections, from mycobacteria to viruses and is therefore not a strong argument for IA if no other documentation is brought, all the more so as coinfections are frequent. This pattern is also associated with a worse functional course with mechanical ventilation requirement ^[6][103].

Both patterns can be intricated in the same patient, showing how complex the IA evolution can be ^[3][100]. Moreover, some patients present with only area of consolidation or ground glass opacities ^[7][8][104,105]. Therefore, whether the CT-scan shows a very specific or a non-specific aspect, further investigations must be performed in search for a mycological criterion for IA. CT-scan is also useful for treatment evaluation.

Guided by CT-scan findings, a bronchoscopy must be performed whenever the patient is in adequate condition (i.e., without severe hypoxia or bleeding) to help better establish the diagnosis through a bronchoalveolar lavage (BAL) ^[1][98]. A higher yield has been obtained in allogeneic stem cell transplant (alloSCT) recipients with new pulmonary infiltrates when bronchoscopy was performed within the first 4 days of presentation ^[9][109]. Saline is usually used for lavage in the segmental or subsegmental bronchus of the most suggestive area identified on the CT-scan. BAL fluid (BALF) must be assessed by direct examination for cytology, Gram and fungal staining. Bacterial and fungal culture must be performed, as well as mycobacterial if relevant. Biomarkers and polymerase chain reaction (PCR) should be assessed as available and clinically relevant: Aspergillus galactomannan (GM), (1,3)-β-D-glucan (BDG), Aspergillus and Mucorales PCR, Pneumocystis jirovecii or Toxoplasma PCR, viral PCR. BAL has a pivotal role in detecting polymicrobial infections, the frequency of which can reach 20% in patients with hematological malignancies and should not be neglected ^[10][11][110,111].

Respiratory samples are diverse and not equal in terms of sensitivity and specificity. While the gold-standard for the diagnosis of IA is bronchoalveolar lavage fluid (BALF), its availability relies on the patient’s fitness and willingness for a not innocuous procedure. Sputum is usually more easily available, but a good quality sample can require physiotherapist maneuvers for patients with a little-productive cough. In intubated patients, tracheal aspirates or brushes are an interesting intermediate sample. However, both sputum and tracheal aspirates may lead to a confusion between infection and colonization if not interpreted with caution.

Culture-positive blood-samples are exceptional, in spite of the angio-invasive capacity of Aspergillus species ^[12][114]. Nevertheless, blood samples are part of the diagnostic work-up of IA, in the search for biomarkers and fungal DNA. According to manufacturer specifications, tests can be performed on whole blood, on plasma or serum. Thresholds for positivity may vary depending on the nature of the sample.

Microscopic examination should be performed on all the available respiratory samples (BALF, tracheal aspiration or sputum), in spite of its low yield. Performance can be enhanced by the use of calcofluor-white ^[13][115]. Aspergillus is identified as a septate hyphae with dichotomous acute angle branching.

Commonly used media for Aspergillus culture are Sabouraud-dextrose, brain-heart-infusion or potato-dextrose agar ^[13][115]. Chloramphenicol is used to avoid competition with bacteria, while cycloheximide can be used to reduce environmental fungal contamination but might reduce the yield. Additional specialized media can be used to select Aspergillus species. Culture should be incubated at 30 °C for 21 days in a humidified environment. Mass spectrometry identification (MALDI-TOF) or genomic sequencing can be used on positive culture if microscopy does not enable a species identification ^[14][15][10,116]. Even more importantly, a positive culture enables testing of the strain’s sensitivity to antifungals.

As IA is associated with immunosuppression, the lack of detection of antibodies directed against Aspergillus cannot be used as a diagnostic argument, contrary to the setting of ABPA or CPA ^[16][117]. Some patients still present with a humoral response leading to the detection of such antibodies within a mean time of 10 days after the onset of IA. The presence of anti-Aspergillus antibodies may decrease the sensitivity of the galactomannan detection test ^[17][118].

GM is a polysaccharide component of the fungal cell wall of Aspergillus. GM detection is validated in serum, BALF and cerebrospinal fluid, with a better sensitivity than culture, and feasible but not validated in sputum and tissue biopsies. The results are expressed as an index in comparison to a control. The cut-off for positivity in serum is still controversial: single value over 0.5 is recommended by the manufacturer, while experts suggest a cut-off over 0.7 for a single test or two consecutive tests > 0.5 ^[18][119]. In BALF too, there is no consensus between >0.5 and >1.0, although a recent meta-analysis including 19 articles concluded in favor of the >0.5 cut-off associated with a sensitivity of 89% and a specificity of 79% ^[19][120].

Its high sensitivity has made GM detection an important part of diagnostic work-up for suspected IA in neutropenic patients. It can reach 60 to 80% in hematology patients but is lower in non-neutropenic patients and poor in the setting of anti-mold prophylaxis ^[17][20][21][118,121,122]. GM can be considered as very specific, but some cross reactivity can occur with other molds or dimorphic fungi such as Fusarium sp., Alternaria sp., Acremonium sp., Penicillium sp., Paecylomyces sp., Wangiella dermatitidis, Histoplasma capsulatum or Blastomyces dermatitidis ^[21][122]. Use of beta-lactam antibiotics or severe mucosal lesions including graft versus host disease (GVHD) have led to false positive tests in serum, while some saline solutions used in BAL have led to false positive tests in BALF.

When positive, an early decrease of serum GM index has been associated with a higher favorable response rate ^[22][123].

Detection of BDG in serum might add sensitivity to the diagnostic work-up ^[23][24][124,125]. However, as this glucan is a component of the cell wall of most fungi, including Aspergillus but also Candida and Pneumocystis, its usefulness is limited by a lack of specificity. As BDG is not produced by Mucorales and Cryptococcus, it can still be of help in differential diagnosis. BDG can be detected also in BALF, but its poor specificity once again limits its usefulness ^[25][126].

Aspergillus PCR has long lacked availability and standardization and is therefore less developed than GM use. These obstacles have now been overcome and PCR is included in routine diagnostic work-up with commercial kits based on serum, plasma or BALF samples ^{[26][27][28][29][30]}[127,128,129,130,131]. Other samples can be analyzed, such as fresh tissue or formalin fixed and paraffin wax embedded tissue ^[31][132].

PCR sensitivity is high and can reach >90% in routine blood samples ^[32][133]. Its combination with GM detection results in BALF has been suggested to reach 96% sensitivity ^[33][134]. PCR specificity is, as expected, very high, up to 100% in some studies ^[34][135]. Cross-reactivities have nevertheless been reported with other molds such as Penicillium spp., Fusarium spp. and Rhizopus oryzae ^[35][136].

An additional value of PCR is the possibility of using probes against resistance-associated mutations, for the detection of azole-resistance, for instance ^[36][137].

Point of care tests, also known as “bedside testing,” are particularly useful for rapid start to relevant antifungal therapy. Two of them available and European conformity CE-marked: lateral flow device AspLFD (OLM Diagnostics, Newcastle-on-Tyne, UK) detecting an extracellular glycoprotein associated with Aspergillus growth, and lateral flow assay (IMMY, Norman, OK, USA) detecting GM ^[37][38][138,139]. Sensitivity and specificity are over 70% ^[39][40][140,141].

Confronted with the heterogeneity of features of IA reported in international literature, experts of the EORTC-MSG (European Organization for the Research and Treatment of cancer/Mycosis Study Group) issued as early as 2002 a system of criteria enabling determination of three levels of diagnostic certainty ^[41][142]. This system was not intended for clinical purpose and did not aim at refraining the start of a treatment due to lack of a criterion. The intent is harmonization of the clinical trials for a better understanding of their results. To account for improvements in diagnostic tools and growing variability of radiological findings associated with IA, updates have been issued in 2008 and 2020 ^[42][43][143,144].

An IA is considered to be proven in the case of a positive mycology or histopathology of a normaly sterile site obtained in a sterile manner. This is restricted to needle aspirations and biopsies as opposed to BAL fluid. Positive mycology can consist of a positive culture, a positive PCR or DNA sequencing. Positive histopathology consists of a sample in which hyphae are seen accompanied by evidence of tissue damage. If an identification of the mold is not performed, the diagnosis will be of IMD rather than IA.

Other levels of diagnostics levels of IA require a more thorough examination of the whole patient’s situation ^[43][144]. The host criterion relies on the presence of a known risk factor among the following: recent neutropenia (neutrophils < 0.500 G/L for more than 10 days), active hematological malignancy, alloSCT, solid organ transplantation, prolonged use of corticosteroids at a minimum dose of 0.3 mg/kg/d of prednisone or equivalent for more than 3 weeks, acute GvHD, T-cell suppressants, B-cell suppressants, inherited severe immunodeficiency. The clinical criterion for pulmonary aspergillosis is met in case of dense, well-circumscribed lesion with or without a halo sign, or an air crescent sign, or a cavity, or a wedge-shaped and segmental or lobar consolidation, whereas the clinical criterion for tracheobronchitis requires a tracheobronchial ulceration, nodule, pseudo-membrane, plaque or eschar seen on bronchoscopy. The mycological criterion is met in case of a positive direct microscopy or culture in sputum, BALF, bronchial brush or aspirate, or GM detected in blood ≥ 1.0, or in BALF ≥ 1.0 or both blood ≥ 0.7 and BALF ≥ 0.8, or a PCR positive in blood twice or in BALF twice (first analysis and duplicate) or once in both blood and BALF.

The diagnosis is probable IA if host criterion, clinical criterion and mycological criterion are all met. It is possible IMD if only host and clinical criterion are met.

In spite of their focus-widening updates, the EORTC-MSG criteria remain restrictive with regard to a large category of patients susceptible to developing IA, particularly in critical care setting. Therefore, parallel diagnostic criteria were developed more adapted to critically ill patients, the Aspergillosis Intensive Care Unit algorithm, known as AspICU ^[44][145]. The proven IA category shares the definition of the EORTC-MSG proven IA. The IA is considered putative in cases in which four criteria are met: the entry criterion is Aspergillus-positive lower respiratory tract specimen culture; the clinical criterion comprises refractory or recrudescent fever despite antibiotics, pleuritic chest pain or rub, hemoptysis, dyspnea or respiratory insufficiency; the radiological criterion consists of an abnormal medical imaging by portable chest X-ray or CT scan of the lungs; the fourth criterion can be either a host risk factor (neutropenia, cytotoxic-treated hematological or oncological malignancy, glucocorticoid treatment or immunodeficiency) or a mycological finding: semiquantitative Aspergillus-positive culture of BALF without bacterial growth together with a positive cytological smear showing branching hyphae. If any of the four criteria is lacking, the case is considered a colonization.

To address the data provided by biomarkers, a new algorithm, BM-AspICU, was proposed ^[45][146]. The entry criterion is either positive Aspergillus in the lower respiratory tract, a radiological sign, or a clinical sign such as described in AspICU. If the patient has a “strong” host factor as defined by EORTC-MSG, a combination of an abnormal imaging and a mycological criterion are sufficient to diagnose a probable IA. If the patient only has a “weak” host factor as defined in AspICU, the combination of a clinical, a radiological and two mycological criteria are needed for the diagnosis of probable IA. Otherwise, the diagnosis is colonization.

More and more often described as a specific entity, influenza-associated pulmonary aspergillosis (IAPA) has its own consensual algorithm for diagnosis, associating pulmonary infiltrates in the setting of PCR-positive influenza with either tracheobronchitis, GM in serum or BALF, culture-positive BALF, or culture-positive sputum associated with an image of cavitation, to draw the diagnosis of probable IAPA ^[46][147].

With the high prevalence of Aspergillus-positive samples in COVID19-patients, a consensus for COVID19-associated invasive aspergillosis (CAPA) had to be found in the literature documenting the epidemy. A system of clinical, radiological and mycological criteria was proposed ^[47][86]. The clinical criteria are similar to that of AspICU: refractory fever, pain, pleuritic rub, dyspnea or hemoptysis. The radiological criterion consists of a pulmonary infiltrate or a cavity. The association of both with a strong mycological criterion (microscopy, culture, GM or PCR positive in BALF) leads to the diagnosis of probable CAPA. Their association with a weak mycological criterion (microscopy, culture, GM or PCR positive in bronchial aspiration) prompts the diagnosis of possible CAPA.

The first antifungal to show activity against Aspergillus was historically the polyene amphotericin B deoxycholate, whose poor safety profile has now led to recommendation against its use ^[48][49][17,148]. It is nevertheless still useful in low-resource settings ^[50][149].

An important improvement came from lipid formulations of amphotericin B, i.e., liposomal amphotericin B (L-amB), amphotericin B lipid complex (ABLC) and colloidal dispersion (ABCD). All were associated with less nephrotoxicity, and none with lower efficacy than amphotericin B  deoxycholate ^[51][52][150,151]. Exhibiting the best safety-profile, L-amB has eventually become the most widely available and used among the lipid formulations of amphotericin B, with a documented efficacy against IA ^[53][16]. The usual dosage against Aspergillus is 3 mg/kg/day and attempts to enhance antifungal activity with higher dosage in highly immunocompromised patients have proven disappointing. L-amB is particularly useful in the setting of a breakthrough infection under azole prophylaxis or in salvage therapy ^[38][139]. Aerosolization of amphotericin B and L-amB can also be used as a prophylaxis for IA, the latter showing a better tolerance ^[54][55][56][152,153,154].

Inhibiting the synthesis of membrane component ergosterol, azoles play a key role in the management of many fungal infections ^[57][155]. Fluconazole lacking any efficacy on Aspergillus, the first azole used in the treatment of IA was itraconazole, with 39% patients having a complete or partial response at the end of treatment but a limiting toxicity and concerns about bioavailability ^[58][156].

A major step in IA treatment was made with the development of voriconazole. In the pivotal randomized trial comparing voriconazole to amphotericin B deoxycholate in IA treatment, successful outcomes reached 52.8% vs. 31.6% ^[48][59][17,157]. Tolerability of voriconazole is generally good, with a restriction due to frequent liver toxicity, visual and neurological side-effects and an interaction profile that may lead to difficulties in managing co-administration with hematological drugs such as cyclosporine, midostaurine or venetoclax. Rarely, prolonged use can lead to phototoxicity, skin carcinomas, peripheral neuropathy and periostitis ^[60][158]. Therapeutic-drug monitoring is available and should be used. The advised dosage is 6 mg/kg bid for a charging period of 24 h followed by 4 mg/kg bid. Voriconazole is recommended as a first-line treatment by IDSA, ECIL and ESCMID ^[1][2][49][98,99,148].

Posaconazole was first used in second line treatment for patients refractory to or intolerant of conventional antifungal therapy, with a satisfactory success rate of 42% ^[61][159]. In terms of tolerance, posaconazole exhibits a better safety profile than voriconazole ^[62][160]. This tolerability has allowed usage as a prophylaxis, all the more so since delayed-release tablets have enhanced bioavailability that used to be a concern ^[63][161]. Another interesting advantage of posaconazole upon voriconazole is its activity against Mucorales ^[64][65][162,163], for which differential diagnosis with IA can be challenging. Direct comparison between voriconazole and posaconazole in the treatment of IA has recently been achieved and shows the absence of inferiority of the latter, with a better safety-profile ^[62][160]. Recommended dosage is 300 mg bid for 1 day, followed by 300 mg qod and therapeutic drug monitoring with a target trough of 0.7 mg/L for prophylaxis or 1.0 mg/L for treatment.

Isavuconazole is the latest approved azole and shows satisfactory activity against Aspergillus, combined with an excellent safety profile. It has been compared with voriconazole in a randomized double-blind study ^[66][164], that demonstrated non-inferiority in terms of efficacy with a similar survival at week 6 and week 12. It was better tolerated with less hepatic, visual and cutaneous adverse events, and can therefore be of particular interest in the treatment of patients at risk of hepatic impairment. Moreover, its interaction profile with various drugs is more easily managed and leads to less dosage adaptations of concomitant therapies. A meta-analysis of isavuconazole trials concluded that its efficacy is comparable to both voriconazole and L-amB ^[67][165]. Besides, isavuconazole shows an interesting activity against Mucorales ^[68][166].

Echinocandins are BDG-synthesis inhibitors that show mitigated activity against IA ^[69][167]. Caspofungin was first used as a second line therapy for IA patients with intolerance or failure of previous therapy and led to an encouraging 45% of favorable responses ^[70][168]. In first line therapy, it has been tried in neutropenic patients and alloSCT patients with results below expectations (53% and 50% survival at week 12, respectively), and is therefore not recommended by guidelines ^[71][72][169,170]. Micafungin and anidulafungin have insufficient activity against Aspergillus to be used as treatment in IA. On the other hand, the upcoming rezafungin shows promising activity against Aspergillus in vitro and mouse models ^[73][171].

Other new antifungals are awaited ^[74][172]. Ibrexafungerp, a BDG-synthesis inhibitor, is currently being tried in combination with voriconazole against voriconazole monotherapy in the setting of IA (NCT03672292; https://clinicaltrials.gov/, accessed on 22 January 2022) after promising results in vitro and mouse models in combination with isavuconazole ^[75][76][173,174].

Olorofim, an inhibitor of dihydro-oroate dehydrogenase, is currently being tried against L-amB in the setting of IA after promising results in vitro and in mouse models of neutropenia ^[77][78][175,176].

Fosmanogepix (APX001), an inhibitor of the fungal enzyme Gwt1, is currently in phase 2 trial after promising preclinical results ^[79][177].

Opelconazole is a new triazole designed for inhalation with a sustained lung residency which might be of use for prophylaxis or treatment of IA ^[74][172].

Several publications have explored the efficacy of combination therapy: L-amB and caspofungin, voriconazole and caspofungin or voriconazole and anidulafungin ^{[80][81][82][83][84]}[178,179,180,181,182]. Only the latter was a prospective randomized large-scale study and no superiority of the combination therapy arm was demonstrated. Therefore, guidelines do not recommend the use of two antifungals for treating IA ^[1][2][49][98,99,148]. However, combination can be argued for in the context of high azole-resistance prevalence and might be of use in salvage therapy.

The availability of many different antifungals with various efficacy and safety profiles prompts the establishment of management strategies ^[85][183]. The relevance of anti-mold prophylaxis for patients at risk after AML-remission induction therapy or during alloSCT procedure must be discussed on the basis of the local epidemiology ^[86][21]. Other situations that can also be an indication for prophylaxis are not as consensual as yet.

Empirical therapy, taking place at the start of antifungal agents in any antibacterial-resistant fever for at-risk neutropenic patients, was proposed four decades ago and decreased morbidity and mortality. However, following studies had conflicting results ^[87][88][184,185]. Concerns are raised about exposing patients to useless toxicities, promoting resistance, and about overall costs.

Preemptive therapy, based on a diagnostic-driven start of antifungals, was made possible through the improvement of efficiency and availability of biomarkers and CT-scan. It has been shown to lessen the consumption of antifungals without impairing patients’ outcome ^[89][90][18,186].

Whichever way the start of treatment is decided, every clinician is faced with the decision regarding the discontinuation of antifungal therapy. Standard recommendations point towards 6 to 12 weeks, but the treatment duration should be tailored to the patient’s situation ^[27][128]. Response assessment with CT-scan, or perhaps PET-CT, can help in the decision-making. If the patient remains at-risk, treatment is often switched into secondary prophylaxis, sometimes with the same antifungal.

An algorithm has been suggested to decide when to stop antifungal treatment in patients with hematological malignancies and IA. It integrates status of the hematological malignancy, recovery from neutropenia, negative mycology, clinical and imaging response to antifungal therapy and planned further chemotherapy and immunosuppression ^[91][108].