Invasive aspergillosis in adult patients in Australia and New Zealand: 2017–2020

Summary

Background

New and emerging risks for invasive aspergillosis (IA) bring the need for contemporary analyses of the epidemiology and outcomes of IA, in order to improve clinical practice.

Methods

The study was a retrospective, multicenter, cohort design of proven and probable IA in adults from 10 Australasian tertiary centres (January 2017–December 2020). Descriptive analyses were used to report patients’ demographics, predisposing factors, mycological characteristics, diagnosis and management. Accelerated failure-time model was employed to determine factor(s) associated with 90-day all-cause mortality (ACM).

Findings

Of 382 IA episodes, 221 (in 221 patients) fulfilled inclusion criteria – 53 proven and 168 probable IA. Median patient age was 61 years (IQR 51–69). Patients with haematologic malignancies (HM) comprised 49.8% of cases. Fifteen patients (6.8%) had no pre-specified immunosuppression and eleven patients (5.0%) had no documented comorbidity. Only 30% of patients had neutropenia. Of 170 isolates identified, 40 (23.5%) were identified as non-Aspergillus fumigatus species complex. Azole-resistance was present in 3/46 (6.5%) of A. fumigatus sensu stricto isolates. Ninety-day ACM was 30.3%. HM (HR 1.90; 95% CI 1.04–3.46, p = 0.036) and ICU admission (HR 4.89; 95% CI 2.93–8.17, p < 0.001) but not neutropenia (HR 1.45; 95% CI 0.88–2.39, p = 0.135) were associated with mortality. Chronic kidney disease was also a significant predictor of death in the HM subgroup (HR 3.94; 95% CI 1.15–13.44, p = 0.028).

Interpretation

IA is identified in high number of patients with mild/no immunosuppression in our study. The relatively high proportion of non-A. fumigatus species complex isolates and 6.5% azole-resistance rate amongst A. fumigatus sensu stricto necessitates accurate species identification and susceptibility testing for optimal patient outcomes.

Funding

This work is unfunded. All authors’ financial disclosures are listed in detail at the end of the manuscript.

Research in context.

Evidence before this study

We reviewed original articles related to epidemiology of invasive aspergillosis. We searched PubMed and Scopus using the terms “invasive aspergillosis OR invasive aspergillus infection”, “retrospective OR prospective”, “epidemiology”. Studies in English until September 2022 were considered and review articles were excluded. Numerous retrospective and prospective studies on epidemiology of invasive aspergillosis have been published, but the majority focused on a particular patient groups, especially those with underlying immunosuppression, such as stem cell or solid organ transplant recipients and patients with haematologic malignancies. There were many single-centre studies, and studies which provided epidemiology and overview of different invasive fungal infections or concurrent infections, whose findings may not be generalisable to our populations. More importantly, there have not been any recent studies evaluating the change in epidemiology of invasive aspergillosis in the Western Pacific region, especially in the current era of new and emerging risk factors for invasive aspergillosis, and rising incidence of azole-resistant Aspergillus fumigatus complex globally.

Added value of this study

Our study provides a contemporary in-depth clinical-epidemiologic analysis of invasive aspergillosis in adult patients in Australia and New Zealand. We observed lower proportion of neutropenic patients in our study cohort, and neutropenia itself was not associated with increased 90-day mortality, potentially due to increasing use of novel, targeted but less toxic therapies in patients with haematologic malignancies. In addition, there were nearly 12% of patients with mild or no immunosuppression in our cohort who developed IA. Notably our study suggested a potentially changing landscape of causative Aspergillus species and susceptibility profile in our region.

Implications of all the available evidence

Invasive aspergillosis is still the most common invasive mould infection in those with traditional host factors, such as those with neutropenia or transplant recipients. However it is now also recognised in patients with mild immunosuppression or no underlying comorbidity. Laboratories should provide accurate identification of Aspergillus to at least species complex level. It is imperative that susceptibility testing of isolates is performed not only for surveillance and monitoring of azole-resistance in the region, but to inform empirical antifungal treatment as well.

Introduction

Invasive aspergillosis (IA) remains a major cause of morbidity and mortality in immunocompromised hosts. Traditional risk factors have included haematological malignancy and transplantation but new risk factors are emerging, including new or more intensive immune-suppressive therapies,^1,2 steroid-dependent chronic lung disease,³ critical illness in an intensive care unit (ICU),⁴ and viral infections most notably influenza⁵ and COVID-19.⁶ Despite modern medical care, global all-cause mortality from IA has remained unchanged and substantial at 35–42%,⁷ with higher mortality rates approaching 60% in patients with relapsed/refractory acute myeloid leukemia.⁸

There have been several large international studies reporting the epidemiology and outcomes of IA, but many were conducted either in specific patient cohorts,^9,10 or were undertaken >10 years ago.^11,12 In Australia and New Zealand, clinical-epidemiological data on IA is limited, with one study focused on length of hospitalisation and costs associated with IA,¹³ and another study describing a range of invasive fungal infections (including IA) in children.¹⁴ Whilst Aspergillus fumigatus species complex is the most common pathogen, species epidemiology varies with geographic regions, with reports of predominance of non-A. fumigatus complex species in the Middle-East,¹⁵ as well as relative high frequency of Aspergillus terreus complex infections in Austria.¹⁶ Likewise, the rate of azole-resistant A. fumigatus is also geography-dependent, with higher resistance rates more frequently reported in Europe (up to 26%)¹⁷ compared with Asia or Middle-East countries (range 1.5%–4.2%)18, 19, 20 or Australia (2.6%).^21,22

To better understand disease patterns of IA, and changes thereof in our region, we present here an in-depth clinical-epidemiologic analysis of IA across Australasia. We describe the predisposing factors, diagnostic methods, causative species, management and outcomes in adult patients with proven and probable IA, and determined predictors of all-cause mortality. The study is timely in view of the recent recognition of A. fumigatus as a “critical pathogen” under the World Health Organisation (WHO) fungal priority pathogen listing.²³

Methods

Study design

This was a retrospective, observational, multicenter study of proven and probable IA in patients $\ge$18 years of age, conducted across 10 tertiary centres in Australia (New South Wales, Queensland, Victoria, Western Australia) and New Zealand (Auckland) from January 2017–December 2020. The study was conducted under the auspices of the Australian and New Zealand Mycoses Interest Group (ANZMIG). All study centres had Human Research Ethics Committee approval (coordinating HREC 2019/ETH11822).

Cases were identified through microbiology laboratory information systems, pharmacy records, and hospital medical records via coding data (ICD-10-AM) for fungal diseases and co-morbidities. Data were collected using a standardized electronic case report form in a secure centralized registry in REDCap (The University of Sydney). Patient demographics, co-morbidities, predisposing factors e.g. immunosuppressive therapies in the preceding 90 days, laboratory and imaging results, antifungal prophylaxis in the past 90 days, and treatment data were captured. For those who were on antifungal prophylaxis in the preceding 90 days, we also identified subset of those who have been on mould-active antifungal prophylaxis for at least 7 day at time of IA diagnosis, where examples of mould-active prophylaxis included any of the tri-azole agents, liposomal amphotericin B and echinocandin. We also regarded ICU admission as a predisposing factor rather an outcome. For patients with >1 episode of IA, only the first episode was included. Patients with possible IA (see Definitions below), mixed invasive mould disease involving a non-Aspergillus mould, allergic forms of aspergillosis and chronic pulmonary aspergillosis²⁴ were excluded.

Definitions

Proven or probable IA was defined according to updated European Organization for Research and Treatment of Cancer-Mycoses Study Group Education and Research Consortium (EORTC-MSGERC) consensus criteria.²⁵ Where cases did not meet these definitions yet had specific predisposing factors such as recent influenza, underlying cardiothoracic transplant, chronic obstructive pulmonary disease (COPD) or ICU admission, other accepted criteria were applied to assign cases as proven or probable IA.3, 4, 5^,26 Each case was reviewed by an adjudication committee (S.C-A.C., M.S., A.J.M., O.M., J.B., B.M., S.Y.T) for study inclusion.

Immunosuppresive therapies were categorized into four subgroups: (i) corticosteroids ($\ge$0.3 mg/kg prednisolone equivalent for $\ge$3 weeks in the past 60 days),²⁵ (ii) conventional cytotoxic cancer chemotherapy, (iii) molecularly targeted therapies which block tumour growth by interfering with specific targeted molecules, such as (but not limited to) checkpoint inhibitors, monoclonal antibodies, immunotherapies, Bcl-2 inhibitor (Venetoclax), and (iv) calcineurin inhibitors (CNI) and disease modifying antirheumatic drugs (DMARDS) [Supplementary Table S1].

Neutropenia and lymphopenia were defined as cell counts $\le$0.5 × 10⁹/L and $\le$1.0 × 10⁹/L, respectively in the preceding 30 days. Results of Aspergillus galactomannan (GM), where measured, were interpreted as ‘positive’ or ‘negative’ according to updated EORTC-MSGERC criteria.²⁵ Disseminated infection involved two or more non-contiguous sites. Primary antifungal treatment was defined as the initial antifungal agent administered for $\ge$7 days. Antifungal therapy was deemed changed if the initial agent was switched to another agent at or beyond day 8.²⁷

Laboratory methods

Clinical specimens, if sent for histopathology, were examined for presence of fungal hyphae with standard Gomori’s methenamine silver, Periodic acid-Schiff and haematoxylin and eosin (H&E) stains.²⁸ All specimens were also examined in the microbiology laboratory by microscopy for fungal hyphae. Culture for fungi was undertaken using standard mycological methods, and fungi were identified by morphological-based methods and, by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS).^29,30 The morphological characteristics of A. fumigatus species complex are well known, with A. fumigatus sensu stricto identified and distinguished by its ability to grow at >50 °C, supported by MALDI-TOF MS identification.²⁹ All non-A. fumigatus species complex aspergilli were identified to species complex level (e.g. Aspergillus flavus complex).

Sequencing of the internal transcribed spacer (ITS) region as well as the β-tubulin gene was performed where species complex/species could not be assigned as above, where identification was uncertain, e.g. for poorly sporulating isolates, and for all non-A. fumigatus sensu stricto A. fumigatus species complex isolates.30, 31, 32 This was undertaken at four reference mycology laboratories (three in Australia, one in New Zealand). Aspergillus PCR and panfungal PCR, where performed was carried out as previously described,^33,34 at a reference laboratory. These centres participate in the Quality Control for Molecular Diagnostics Aspergillus spp. DNA EQA program (QCMD; Glasgow, U.K.) for Aspergillus PCR and sample exchange programs for proficiency testing. Aspergillus GM levels were measured using the Platelia™ Aspergillus antigen assay (Bio-Rad, Marnes-la-Coquette, France).

Antifungal susceptibility testing was performed using a Sensititre® YeastOne™ YO10 panel (TREK Diagnostics, Cleveland, OH), this panel is registered in Australia and New Zealand by the Therapeutics Goods Administration for susceptibility testing of Apsergillus spp. and has good agreement with the Clinical Laboratory Standards Institute (CLSI) standard for Aspergillus spp.35, 36, 37, 38, 39, 40, 41 Minimum inhibitory concentrations (MICs; for amphotericin B, itraconazole, voriconazole, posaconazole) and minimum effective concentrations (MECs; for echinocandins) were interpreted using Clinical Laboratory Standards Institute (CLSI) clinical breakpoints (CBPs) or where there are none, CLSI epidemiological cutoff values (ECVs) to define wild-type (WT) or non-WT isolates.^40,41

Radiology

For pulmonary computed tomography (CT) scans, each radiology report was reviewed for specific abnormalities according to published criteria including halo sign, air crescent sign, dense, circumscribed lesions (nodules), consolidation, single or multiple cavities.²⁵ Findings such as non-specific pulmonary or parenchymal infiltrates (e.g. hazy airspace opacification), which were not part of the revised definitions of the radiologic criteria of lower respiratory tract invasive fungal disease were also recorded, as were abnormalities in non-pulmonary radiological images.

Outcome

Patient outcome data were collected at 30-, 90- and 180-days from the time of IA diagnosis, or until death (whichever occurred sooner). The primary outcome was 90-day all-cause mortality, with 30-day and 180-day all-cause mortality as secondary outcomes. Death attributed to IA was recorded.

Statistical analysis

Descriptive analyses were used to report demographics, co-morbidities, predisposing factors, diagnostic methods and management. Accelerated failure-time analyses were employed to ascertain factor(s) associated with 90-day all-cause mortality from IA diagnosis, followed by Kaplan–Meier survival analyses. Covariates with p < 0.1 in univariate analysis were introduced to the multivariate analysis. The multivariate analysis was limited to a maximum number of concurrent predictors based on number of patients who died to avoid overfitting. Description of patient characteristics and outcomes was also performed for the following subpopulations: patients with haematologic malignancies (HM), solid organ transplant (SOT) recipients and those without comorbidity, with accelerated failure-time model performed for primary outcome analysis in the HM subgroup (given the higher proportion in the study cohort). Statistics were performed with Stata version 16.1 (StataCorp LP, College Station, Texas, US), with p-values < 0.05 considered significant.

Role of the funding source

This work is unfunded. All authors’ financial disclosures are listed in detail at the end of the manuscript.

Results

Of 382 episodes of infection, 221 cases (in 221 patients) fulfilled the inclusion criteria; 161 episodes were excluded because they failed to meet host criteria (n = 19) or microbiological definitions (n = 39), did not have compatible radiological findings (n = 42), or failed to meet a combination of these criteria (n = 39). One case had mixed Aspergillus-Scedosporium infection and 21 were paediatric cases. Hence we evaluated 53 proven (one diagnosed post-mortem), and 168 probable IA infections.

Patient characteristics

The median patient age was 61 years [interquartile range (IQR) 51–69]; 136 (61.5%) were males. Patients with underlying HM comprised 49.8% of patients (n = 110), followed by SOT recipients (n = 55, 24.9%). Fifteen (6.8%) did not have the pre-specified host criteria of immunosuppression but had primary comorbidity as follows: eight with chronic obstructive pulmonary disease, four with diabetes mellitus, two with chronic liver disease and one with chronic renal impairment. Eleven patients (5.0%) had no co-morbidities though some had predisposing factors: two patients had preceding trauma, one with endovascular aortic surgery 40 years prior and one had received a prolonged course of corticosteroids (reason not stated) (see Table 1 and Supplementary Table S2).

Table 1.

Patient demographics and co-morbidities [all percentages are calculated using the total number of cases (221) as denominator].

Entire patient cohort	Results (n, %)
Total number of cases	221 cases
Comorbiditiesa
Haematological malignancies	110 (49.8%)
•Leukemia	63 (28.5%)
•Lymphoma	29 (13.1%)
•Myeloma	5 (2.3%)
•Others (myelodysplasia, myeloproliferative disorder, aplastic anaemia, HLH, autoimmune lymphoproliferative disorder, plasmacytoid dendritic cell neoplasm)	13 (5.9%)
Haematopoietic stem cell transplant (HSCT)b	38 (17.2%)
•Autologous	11 (5.0%)
•Allogeneic	27 (12.2%)
Solid organ transplant	55 (24.9%)
•Heart	5 (2.3%)
•Liver	6 (2.7%)
•Lung	15 (6.8%)
•Kidney	28 (12.7%)
•Concurrent lung and kidney	1 (0.4%)
Solid organ malignancy	28 (12.7%)
Primary immunodeficiency	4 (1.8%)
Chronic lung diseasec	55 (24.9%)
Autoimmune disordersd	19 (8.6%)
Diabetes mellituse	43 (19.5%)
Chronic liver diseasef	7 (3.2%)
Chronic kidney diseaseg	28 (12.7%)
No comorbidity	11 (5.0%)
Predisposing factors
Therapies
a) Corticosteroid useh	53 (24.0%)
• Alone	13 (5.9%)
b) Targeted therapiesi	57 (25.8%)
• Alone	15 (6.8%)
a) Conventional cancer chemotherapies	70 (31.7%)
• Alone	31 (14.0%)
b) Calcineurin inhibitors/DMARDs	79 (35.7%)
• Alone	50 (22.6%)
Not on any of these therapies	44 (19.9%)
On a combination of drug therapies	68 (30.8%)
Patients with neutropenia in preceding 30 daysj	66 (29.9%)
Patients with lymphopenia in preceding 30 daysk, l	115 (54.8%)
Previous trauma or non-transplant surgery	3 (1.3%)
Intensive care unit (ICU) admission	47 (21.3%)
Median length of stay in ICU (days)	9 (IQR 2–20)

HLH = haemophagocytic lymphohistiocytosis; IA = invasive aspergillosis; DMARDs = disease modifying anti-rheumatic drugs; IQR = interquartile range; COPD = chronic obstructive pulmonary disease.

^aEach patient might have more than one comorbidity; altogether 98 patients had two or more comorbidities.

^bThe indication for all the HSCT procedures was haematologic malignancies.

^cChronic lung diseases included: predominantly COPD, followed by asthma, mycobacterial (tuberculous and non-tuberculous) infections, cystic fibrosis, bronchiectasis, chronic pulmonary graft-versus-host-disease, alpha-1-antitrypsin deficiency and autoimmune pulmonary conditions. Forty-seven patients had concurrent chronic lung disease with other forms of immunosuppression, and eight had COPD only.

^dAutoimmune diseases included: systemic lupus erythematosus, ANCA vasculitis, Sjogren’s syndrome, polyarteritis nodosa, granulomatosis with polyangiitis, cryoglobulinemic vasculitis, mixed connective tissue disease, rheumatoid arthritis, IgG4 disease, polymyalgia rheumatica and scleroderma.

^eFour of 43 patients had diabetes mellitus as their only comorbidity.

^fTwo of seven patients had chronic liver disease as their only comorbidity.

^gOne of 28 patients had chronic renal impairment as their only comorbidity.

^hCorticosteroid use was defined as a dose of ≥0.3 mg/kg prednisolone equivalent for ≥3 weeks in the preceding 60 days of IA diagnosis.

ⁱTargeted therapies included checkpoint inhibitors, monoclonal antibodies, immunotherapies, newer therapies such as Bcl-2 inhibitor (Venetoclax) (See supplementary information Supplementary Table S1 for a comprehensive list of therapies recorded).

^jNeutropenia is defined as neutrophil count $\le$ 0.5 × 10⁹/L.

^kLymphopenia is defined as lymphocyte count $\le$ 1.0 × 10⁹/L.

^lEleven patients’ lymphocyte counts were not captured, hence the denominator was only 210.

Overall, 177 patients (80.1%) received one of the following therapies: corticosteroids, conventional chemotherapies, targeted therapies or CNI/DMARDS; overall 68 patients (30.8%) were on a combination of the above (see Table 1 and Supplementary Table S1). Conventional chemotherapy and targeted therapies were prescribed predominantly for patients with HM: 91.4% (n = 64) and 86.0% (n = 49), respectively. CNI/DMARDS were most frequently prescribed in non-HM patients (n = 57, 72.1%). Sixty-six patients (29.9%) had neutropenia for a median of 14 days (IQR 8–30), of which 62 (93.9%) had underlying HM. Lymphocyte count was available for 210 patients, of these 54.8% had lymphopenia (n = 115).

Ten patients (4.5%) had antecedent or concurrent influenza, of which 5 were admitted into ICU and 3 needed mechanical ventilation. The median time from influenza diagnosis to IA was 6 days (range 0–18). Overall 47 patients (21.3%) required ICU admission, with lengthier median length of stay for those who died (10 versus 5 days).

Fifty-one patients (23.1%), majority of whom had underlying HM (96.0%), received yeast or mould antifungal prophylaxis in the preceding 90 days. Of these, 20 patients (47.1%) had been on mould-active antifungal prophylaxis for at least 7 day at time of IA diagnosis. For these 20 patients, posaconazole was most commonly prescribed (n = 15, 75.0%), with a median duration of prophylaxis of 23 days (IQR 14–110) prior to diagnosis of IA, followed by voriconazole [n = 4, 20.0%, median duration 60 days (IQR 44–338)] and liposomal amphotericin B [n = 2, 13.7%, median duration 18 days (IQR 12–25)].

Site of infection

Of 221 cases, eight were disseminated infections. Localised infections included invasive pulmonary aspergillosis (IPA) (n = 181), followed by sino-nasal disease (n = 15), infection affecting bone/joint (n = 6), central nervous system (CNS) infections (n = 2), skin/soft tissue (n = 2), genitourinary system (n = 2), eye (n = 2), blood vessels, aortic graft and pacemaker (each n = 1). Body sites implicated in disseminated infections were: pulmonary with CNS (n = 4), combined sino-pulmonary infections (n = 2), pulmonary with skin/soft tissue (n = 1), and pulmonary with spine and oesophagus (n = 1).

Diagnosis

Of 53 proven cases, histopathological examination identified hyphae typical of Aspergillus spp. with associated necrosis and tissue damage in 30 patients (56.6%), and Aspergillus spp. were cultured from 42 cases (79.2%). Indirect tests were also positive in 20 of these cases (37.7% - positive Aspergillus PCR, n = 15; positive GM, n = 5) (Supplementary Table S3).

For probable cases, fungal elements were identified by microscopy in 46 non-sterile samples (27.4%). While 111 probable cases (66.1%) had positive Aspergillus cultures, 57 cases (33.9%) were diagnosed solely by indirect methods [serum GM only (n = 5), positive GM in broncho-alveolar lavage only (n = 23), Aspergillus PCR only (in various samples e.g. vitreous fluid, BAL, blood, n = 6), positive GM and Aspergillus PCR in BAL (n = 22), positive serum GM and Aspergillus PCR in BAL (n = 1)] (Supplementary Table S3).

With regards to overall performance of GM, there were 24 patients who had serum GM performed. Of these 11 had optical density index (ODI) of $\ge$1.0 (45.8%), eight with negative result according to the EORTC/MSGERC criteria (33.3%) – where four had ODI between 0.5–0.99 and four with ODI < 0.5; whereas 89 patients had BAL GM performed, of which 76 had ODI of $\ge$1.0 (85.4%), and eight patients’ BAL GM did not meet the EORTC/MSGERC criteria (9.0%)—7 of them with ODI between 0.5–0.99, and one patient had ODI of 0.1. The remaining five patients (from each group) had both serum ODI $\ge$ 0.7 and BAL fluid ODI $\ge$ 0.8, positive according to the mycological criteria by EORTC/MSGERC.

Overall, the median ODI serum and BAL GM were 1.18 (IQR 0.5–2.01) and 3.67 (IQR 1.56–5.76), respectively.

Causative species

Of 221 cases, causative pathogens were identified in 158, yielding 170 causative pathogens–165 Aspergillus fungi were cultured from 153 cases and in 5 cases, the pathogen was identified by panfungal PCR/DNA sequencing alone. One hundred and forty-two patients were infected with a single Aspergillus pathogen, ten had mixed infection with two pathogens and one patient had three different aspergilli cultured from bone (Fig. 1).

Fig. 1.

Number of patients with causative Aspergillus species identified and methods of identification.

A. fumigatus sensu stricto was the commonest causative pathogen (n = 128/221, 57.9%) accounting for 81.0% (128/158) of instances where species/species complex could be assigned (Fig. 1), followed by Aspergillus niger complex (n = 14, 8.9%) and Aspergillus flavus complex and Aspergillus terreus complex (each, n = 10, 6.3%). Other species within Section Fumigati including cryptic species of the A. fumigatus species complex [Aspergillus lentulus, Aspergillus fischeri; (each n = 1)], and Aspergillus felis (n = 1) were rare. Overall, non-A. fumigatus species complex aspergilli comprised 40/170 (23.5%) of aspergilli identified in specimens and accounted for infection in 38 patients (24.0%) where the causative pathogen species was identified.

Antifungal suscetibilites

The MICs values for seven antifungal agents of 62 Aspergillus isolates are given in Supplementary Table S4. The MIC₅₀, MIC₉₀, geometric mean MIC values and MIC ranges for 46 A. fumigatus sensu stricto isolates are summarised in Table 2, as are percentage of isolates that were assigned as ‘resistant’ or non-WT.^40,41 MIC ranges for other species complexes where more than one isolate was tested are shown in Table 3.

Table 2.

The in vitro susceptibilities of 46 Aspergillus fumigatus sensu stricto isolates to seven antifungal agents.

Pathogen (number of isolates)	MIC or MECe values (mg/L)
	MIC range	MIC50	MIC90	Geometric mean MIC	% resistant (number)a	% non-WT (number)b
Aspergillus fumigatus sensu stricto (n = 46)
Amphotericin B	1–4	2	2	1.86	-	10.9% (5)
Itraconazole	0.12–>16	0.25	1	0.39	-c	6.5% (3)
Voriconazole	0.25–8	0.25	1	0.44	4.3% (2)d	4.3% (2)
Posaconazole	0.12–1.0	0.12	0.5	0.18	-	-
Caspofungin (n = 32)e	0.015–0.06	0.03	0.06	0.04	-	0% (0)
Anidulafungine	<0.015–0.025	0.015	0.03	0.02	-	-
Micafungine	<0.008–0.06	0.008	0.015	0.01	-	-

MEC = minimum effective concentration; MIC = minimum inhibitory concentration; non-WT = non-wild type; - = not relevant.

^aThe clinical breakpoint for A. fumigatus sensu stricto for voriconazole is $\le$0.5 mg/L.

^bThe epidemiological cut off values for amphotericin B, itraconazole, voriconazole and caspofungin for A. fumigatus sensu stricto are 2, 1, 1 and 0.5 mg/L, respectively.

^cHowever 3 strains had itraconazole MICs of >16 mg/L, hence would caution against using this agent.

^dIn total there were four isolates with voriconazole MIC >0.5 mg/L: two isolates with MIC of 1 mg/L (which made them intermediate in terms of susceptibility to voriconazole), two isolates with MIC of 8 mg/L (resistant) (see Supplementary Table S4).

^eThe minimum effective concentrations (MEC) only applies to echinocandins (caspofungin, anidulafungin and micafungin).

Table 3.

MIC ranges of Aspergillus niger complex, Aspergillus flavus complex and Aspergillus terreus complex to seven antifungal agents.

Pathogen (number of isolates)	MIC or MECa values (mg/L)
	MIC range	% non-WT (number)
Aspergillus niger complex (n = 4)
Amphotericin B	0.5–2.0	0% (0)
Itraconazole	0.5–1.0	0% (0)
Voriconazole	0.5–1.0	0% (0)
Posaconazole	0.06–0.5	0% (0)
Caspofungin (n = 2)a	0.015–0.06	0% (0)
Anidulafungina	0.015–0.03	-
Micafungina	0.008–0.03	-
Aspergillus flavus complex (n = 4)
Amphotericin B	4–8	25.0% (1)
Itraconazole	0.06–0.5	0% (0)
Voriconazole	0.5–1.0	0% (0)
Posaconazole	0.06–0.5	0% (0)
Caspofungin (n = 4)a	0.03–0.5	0% (0)
Anidulafungina	0.015–0.25	-
Micafungina	0.015–0.5	-
Aspergillus terreus complex (n = 4)
Amphotericin B	2–4	0% (0)
Itraconazole	0.25–0.5	0% (0)
Voriconazole	0.25–0.5	0% (0)
Posaconazole	0.25–0.5	0% (0)
Caspofungin (n = 3)a	0.03–0.12	0% (0)
Anidulafungina	0.03–0.12	-
Micafungina	<0.008–0.16	-

MEC = minimum effective concentration; MIC = minimum inhibitory concentration; non-WT = non-wild type; - = not relevant.

^aThe minimum effective concentrations (MEC) only applies to echinocandins (caspofungin, anidulafungin and micafungin).

Briefly, geometric mean MICs of itraconazole and voriconazole for A. fumigatus sensu stricto strains were similar (0.39 and 0.44 mg/L, respectively) whilst that of posaconazole was about 2-fold lower (0.18 mg/L). MIC₉₀ values (1 mg/L) for itraconazole and voriconazole were at the ECVs for these agents.⁴¹

Three (6.5%) isolates were azole-resistant: two (isolates RMH-100 and strain 3528-28, see Supplementary Table S4) displayed high MICs to itraconazole, voriconazole and posaconazole with MICs of >16, 8 and 1 mg/L and >16, 8, and 2 mg/L respectively. The remaining A. fumigatus sensu stricto isolate (Westmead-021) had high MICs of >16 and 8 mg/L against itraconazole and posaconazole respectively, but was susceptible to voriconazole. Two isolates (Westmead-030 and 3528–40) had voriconazole MIC of 1 mg/L, which rendered them intermediate in terms of clinical breakpoint for voriconazole, but were still considered WT based on ECV for voriconazole. In addition, five isolates (10.9%) demonstrated amphotericin-B MICs of 4 mg/L, above the ECV (2 mg/L) for this drug–species combination; four of these had WT MICs to itraconazole, low MICs to posaconazole and were not resistant to voriconazole except for isolate 3528-28 as mentioned above.

MECs values for caspofungin were only available for 44 (71.0%) isolates. However, in general, geometric mean MECs were low for the echinocandins (<0.5 mg/L) with micafungin demonstrating the greatest activity (geometric mean MIC 0.01 mg/L; Table 2).

Both A. fischeri and A. felis showed relatively high MICs to voriconazole (both 2 mg/L, Supplementary Table S4). Non-A. fumigatus complex species had 2–4 hold higher amphoterin-B MICs compared with A. fumigatus species complex, with the single Aspergillus versicolor complex (8 mg/L), A. terreus complex and A. flavus complex (both 4 mg/L) having the highest amphotericin-B MICs whilst azole MICs were broadly similar. Amongst 12 A. niger, A. terreus, and A. flavus complex isolates, non-WT MICs were very uncommon, with only one A. flavus complex being non-WT to amphotericin-B (Supplementary Table S4).

We examined the clinical details of three patients with azole-resistant isolates, and found that two had been on mould-active antifungal prophylaxis for at least two years prior to diagnosis, and the third patient was a smoker with COPD with proven IA diagnosed post right upper lobe lobectomy and had no prior exposure to antifungal prophylaxis.

Radiological features

Overall 186 patients (84.2%) with IPA had abnormalities on chest CT. Specific abnormalities according to published criteria²⁵ were present in 183 patients (98.4%), with 57 (31.1%) also having atypical findings such as non-specific pulmonary/parenchymal infiltrates, fungal balls, and pleural thickening (Supplementary Fig. S1).

Pulmonary nodule(s) was the most frequent abnormality (n = 124), and patients with pulmonary nodules and/or consolidation comprised 92.5% (n = 172) of abnormal CT chest findings. Halo and air crescent signs were not common in our study population – only present in 10.2% (n = 19/186, where 11 patients were neutropenic) and 2.2% (n = 4/186) of cases, respectively.

In the SOT cohort, presence of consolidation was the most frequent finding, (n = 24/55, 43.6%), followed by pulmonary nodules (n = 19/55, 34.5%).

Management

Treatment was initiated in 218 patients (98.6%): 187 received antifungal treatment alone (85.8%), 3 underwent surgery alone (1.4%), and 28 received antifungal drugs plus adjunctive surgery (12.8%). Of 215 patients on antifungal treatment, 201 continued the primary antifungal regimen as 14 died and received <7 days of antifungals. Voriconazole monotherapy was the most frequently prescribed primary antifungal (n = 134, 72.0%) (Table 4). Fifteen patients (7.5%) received combination therapy. Sixty-two patients (30.8%) switched to alternative antifungal therapy after a median of 22 days (IQR 11–62).

Table 4.

Proportion of 201 patients on different primary antifungal therapy and median duration of treatment for each monotherapy.

Patients on monotherapy as their primary antifungal regimen (total n = 186)
	Number (%)	Median duration, days (IQR)
Voriconazole	134 (72.0%)	80 (IQR 21–128)a
Posaconazole	26 (14.0%)	86 (IQR 17–180)b
Itraconazole	3 (1.6%)	93c
Liposomal amphotericin B	19 (10.2%)	21 (12–34)
Echinocandin	3 (1.6%)	13 (7–29)
Others (fluconazole)	1 (0.5%)	18
Patients on combination of therapies as their primary antifungal regimen (total n = 15) Voriconazole + echinocandin (n = 4) Voriconazole + echinocandin + nebulised amphotericin B (n = 2) Liposomal amphotericin B + voriconazole (n = 5) Liposomal amphotericin B + posaconazole (n = 1) Liposomal amphotericin B + echinocandin (n = 2) Posaconazole + echinocandin (n = 1)
Patients who underwent surgical intervention (total n = 31) Lobectomy/wedge resection/thoracotomy (n = 9) Extensive surgical debridement (e.g. septoplasty, maxillectomy, ethmoidectomy, uncinectomy) (n = 13) Drainage of brain abscesses (n = 3) Others (n = 6)d

^aNine patients’ duration of treatment for voriconazole was not available.

^bFour patients’ duration of treatment for posaconazole was not available.

^cTwo patients’ duration of treatment for itraconazole was not available.

^dOther surgeries included right common iliac stent explantation and femoral–femoral crossover graft with cadaveric iliac artery (n = 1), left tarsorrhaphy, left division tarsorrhaphy and evisceration (n = 1), pacemaker removal (n = 1), sternal wound debridement and sternal wire removal (n = 1), insertion of oesophageal stent (n = 1), prostatectomy (n = 1).

Outcomes

All-cause 90-day mortality was 30.3% (n = 67), with 16/67 (23.9%) deaths attributed to IA. Median time from diagnosis of IA to death was 22 days (IQR 11–49). All-cause 30-day and 180-day mortality was 18.1% (n = 40) and 35.3% (n = 78), respectively.

Of note, 90-day mortality was similar for the probable IA cases diagnosed by culture compared with those diagnosed by indirect methods alone (41/111, 36.9% vs. 19/57, 33.3%). Table 5 summarizes the factors associated with mortality. Multivariate analysis identified only HM (HR 1.90; 95% CI 1.04–3.46, p = 0.036) and ICU admission (HR 4.89; 95% CI 2.93–8.17, p < 0.001) in predicting mortality (Fig. 2), but not solid organ transplantation (HR 0.65; 95% CI 0.35–1.20, p = 0.178) or presence of neutropenia (HR 1.45; 95% CI 0.88–2.39, p = 0.135).

Table 5.

Univariate and multivariate analysis of all-cause 90-day mortality for a) the entire cohort^a; b) for the haematologic malignancy cohort.

a) For the entire cohorta
Variable	Univariate analysis				Multivariate analysis
	Hazards ratio	95% CI		p-value	Hazards ratio	95% CI		p-value
Age	1.01	0.99	1.02	0.181
Sex	0.84	0.51	1.39	0.520
Haematologic malignancies	1.86	1.13	3.06	0.014	1.90	1.04	3.46	0.036
HSCT	1.34	0.74	2.43	0.319
Solid organ transplant	0.65	0.35	1.20	0.178
Solid organ malignancies	1.10	0.54	2.22	0.784
Autoimmune disorder	1.39	0.63	3.04	0.407
Chronic lung disease	1.50	0.90	2.52	0.119
Diabetes	1.15	0.64	2.08	0.624
Chronic kidney disease	1.41	0.74	2.69	0.295
Chronic liver disease	2.26	0.82	6.21	0.113
No comorbidity	0.25	0.03	1.81	0.171
Influenza	2.70	1.23	5.92	0.013	1.83	0.78	4.33	0.163
Neutropenia	1.45	0.88	2.39	0.135
Lymphopenia	2.13	1.25	3.61	0.005	1.29	0.71	2.37	0.395
Disseminated disease	1.75	0.55	5.58	0.342
ICU admission	5.51	3.40	8.95	<0.001	4.89	2.93	8.17	<0.001

b) For the haematologic malignancy cohort
Variable	Univariate analysis				Multivariate analysis
	Hazards ratio	95% CI		p-value	Hazards ratio	95% CI		p-value
Age	1.00	0.98	1.02	0.807
Sex	1.23	0.65	2.31	0.518
HSCT	0.94	0.49	1.79	0.860
Autoimmune disorder	2.65	0.64	10.99	0.178
Chronic lung disease	1.69	0.86	3.30	0.124
Diabetes	2.21	1.02	4.80	0.043	1.82	0.73	4.57	0.196
Chronic kidney disease	5.08	1.79	14.39	0.002	3.94	1.15	13.44	0.028
Influenza	1.09	0.15	7.94	0.931
Corticosteroid (either alone or in combination with other therapies)	2.01	1.08	3.73	0.026	0.87	0.43	1.75	0.710
Targeted therapies only	0.60	0.21	1.70	0.343
Cancer chemotherapy only	0.43	0.18	1.03	0.060
Neutropenia	1.14	0.61	2.11	0.668
Lymphopenia	1.42	0.65	3.09	0.371
Disseminated disease	2.23	0.54	9.25	0.266
ICU admission	5.84	3.15	10.83	<0.001	6.44	3.29	12.57	<0.001

HSCT = haemopoietic stem cell transplant.

Statistically significant values (p < 0.05) are indicated in bold.

^aIn the final model, there were only 209 patients available for multivariate analysis, as one patient’s date of death was not captured, and 11 patients’ lymphocyte count was not recorded.

Fig. 2.

Kaplan–Meier survival curve stratified by: a) presence or absence of haematologic malignancy for the entire cohort (total 220 patients, as one patient’s date of death was not captured); b) ICU admission for the entire cohort (total 220 patients, as one patient’s date of death was not captured); c) presence or absence of chronic kidney disease for the haematologic malignancy subgroup only (total 110 patients); d) ICU admission for the haematologic malignancy subgroup only (total 110 patients).

Patient subgroup analyses

a) HM patients (n = 110)

All-cause 90-day mortality was 38.2% (n = 42), with 23.8% of deaths (n = 10) attributed to IA in this subgroup. All-cause 30-day and 180-day mortality were 20.9% (n = 23) and 45.4% (n = 50) respectively. Univariate and multivariate analysis demonstrated concurrent chronic kidney disease (HR 3.94; 95% CI 1.15–13.44, p = 0.028) and ICU admission (HR 6.44; 95% CI 3.29–12.57, p < 0.001) to be significantly associated with 90-day mortality (Table 5).

b) Organ transplant recipients (n = 55)

All-cause 90-day mortality was 23.6% (n = 13) and the 30-day and 180-day mortality were 12.7% (n = 7) and 27.3% (n = 15), respectively; 23.1% of deaths (n = 3) were abscribed to IA. Overall median duration from SOT to diagnosis of IA was 260 days (IQR 70–1101). For lung and renal transplant recipients specifically, median times after transplantation to IA diagnosis were 304 days (IQR 38–1512), and 418 days (IQR 115–1308), respectively.

c) Patients without comorbidity (n = 11)

Supplementary Table S2 summarizes the clinical characteristics of the 11 patients diagnosed with IA (nine proven and two probable cases), but without documented comorbidities apart from two with preceding trauma and one with prolonged course of corticosteroids before diagnosis of IA. There was only one death in this subgroup (proven Aspergillus aortitis).

Discussion

This is one of the largest, multicenter in-depth clinical-epidemiologic evaluation of adult IA cases in the Southern Hemisphere. Our results highlight the following key findings: a) targeted therapies were increasingly used in the HM population, with neutropenia being less common in patients with IA, and itself not being associated with mortality; b) a high proportion of hosts with mild or no immunosuppression (12%) c) causative Aspergillus species are diverse, with 6.5% rate of azole-resistance detected amongst 46 tested A. fumigatus isolates.

HM has remained the most highly represented group in our study (49.8%), and a significant contributor to death (p = 0.036), in keeping with the high risk profile of these patients. We observed a myriad of HM treatment protocols involving targeted agents increasingly used in this subgroup (Supplementary Table S1). Novel targeted therapies such as Bruton tyrosine kinase (BTK) inhibitors, BCL-2 inhibitor, checkpoint inhibitors and monoclonal antibodies are known to improve overall survival for the underlying HM, yet their association with death related to IA remains ill-defined.^42,43 Indeed, our analysis demonstrated that these therapies were not significantly associated with mortality. However, the diversity of immunosuppression protocols and dosing regimens limited our confidence in this finding, hence post-marketing surveillance of IFD rates with these new therapies is of paramount importance.

Our study demonstrated an overall 90-day mortality of 30.3% in patients with IA: slightly lower compared to earlier publications with similar patient profiles (36–50%),^7,11,12 likely reflecting improved diagnostic and management strategies in the modern era. In contrast to a recent study by Bretagne et al.⁷ which reported 3-month mortality of 42.5%, with higher global mortality in cases with concurrent infections, we excluded patients with mixed invasive mould disease involving a non-Aspergillus mould due to our intention to evaluate mortality associated with IA alone. In an adjusted analysis of both the entire cohort and HM subgroup, ICU admission was a significant predictor of mortality in patients with IA, with non-survivors experiencing longer length of ICU stay, reflecting a seriously ill cohort of patients with a poorer prognosis. In the HM subgroup, patients with concurrent chronic kidney disease had increased mortality, consistent with previous reports demonstrating renal impairment to be a risk factor for developing IA, along with being a prognostic indicator.^11,44

Notably, in contrast to other IA studies,^11,45 neutropenia was not significantly associated with mortality. In addition to inadequate sample size, this may also be due to the expanding role of novel, targeted and less toxic treatment protocols causing less frequent neutropenia in HM (occurred in only 56.4% of all HM patients in this series). Recent reports of newer regimens, such as venetoclax (in combination with hypomethylating agents) in patients with de novo acute myeloid leukemia, indicate protracted neutropenia without an increased risk of IFD. It has been postulated that these new targeted agents may be associated with less mucosal disruption than intensive chemotherapy, which accordingly may reduce risk of IFD.^42,43 Further observation is needed regarding this field of rapidly expanding treatment regimens for various HMs. Although mortality associated with IA remains significantly high in patients with relapsed or refractory HM or unresolved neutropenia,^8,42,45 we did not observe a large number of such patients in this study.

IA in SOT is associated with 20–46% of graft loss⁴⁶ and mortality up to 25%,⁴⁷ compatible with the mortality rate we observed. Our finding of late-onset IA following SOT is also consistent with previous studies. IA was diagnosed a median of 7.7 months post lung transplantation in a large international cohort study 4 4⁴⁸; and a mean interval of 2.8 ± 3.7 years and median of 34.4 months in two studies of renal transplant recipients.^49,50 These results have direct implications for prophylactic strategies and clinical management when SOT patients present late with pulmonary infiltrates.

Our study also identified development of IA in immunocompetent hosts without co-morbidities or those with mild immunosuppression or chronic conditions (11.8%). Importantly, amongst the 140 cases excluded post-adjudication for not meeting the proven or probable definitions, 50 cases were deemed clinically significant, of which 15 cases lacked pre-defined host criteria and were managed by the treating physician with antifungal drugs, with clinical response. This highlights one limitation of the published consensus criteria in diagnosing IFD. In a clinical trial context, these 50 patients would not have been eligible for inclusion. Concerted international efforts are therefore needed to refine and validate the definitions of IFD in patient groups without classical immunocompromised conditions.

Whilst A. fumigatus sensu stricto remains the commonest causative pathogen as consistent with reports elsewhere,⁵¹ species outside of the A. fumigatus complex are increasingly recognised as pathogens causing IA. Herein 23% of all 170 Aspergillus isolates were non-A. fumigatus complex species, compared with rates of 12–36% reported elsewhere.52, 53, 54 As demonstrated in our study, these species can exhibit intrinsic or acquired resistance to antifungal agents,⁵⁵ highlighting the need moving forward, for accurate species identification and susceptibility testing of these species to guide clinical management.

The global rise in rates of azole-resistant A. fumigatus is driven predominantly by widespread use of azole fungicides in agriculture, and to a lesser extent, selection pressure after prolonged treatments in chronically infected patients,⁵⁶ as supported by findings in two of our patients. We detected here a 6.5% rate of azole-resistance amongst the 46 isolates of A. fumigatus sensu stricto; this result may be biased as only 46/128 (36%) of isolates were tested and prospective systemic studies are required to accurately assess the rates of azole resistance. Two previous Australian studies have reported an azole-resistance rate of approximately 2.6%, amongst clinical A. fumigatus sensu stricto isolates.^21,22 This contrasts with the results in some European countries, with rates as high as 20% amongst clinically significant A. fumigatus sensu stricto isolates.⁵⁷ Nonetheless our observation still calls for surveillance for azole-resistance amongst human isolates, and those from the environment, to more accurately determine the burden of azole-resistance and non-wild type isolates in our region. Of interest, 11% of A. fumigatus sensu stricto isolates had non-WT MICs to amphotericin-B; whilst they could not be classed as exhibiting resistance, and there are no data linking patient outcomes with amphotericin B MICs above the ECVs,^58,59 such findings indicate the need for mechanistic studies of resistance and add support for the case of systematic surveillance for drug resistance.

Our study was limited by its retrospective nature and possible ascertainment bias. We did not have many heart, lungs and liver transplant recipients in our study, hence results may not be applicable to these cohorts. The study also lacked data related to COVID-19 associated pulmonary aspergillosis (CAPA) as it was infrequently observed in Australasia during study period (only one case report published online by end of 2020) due to quarantine and border controls. We were not able to assess rates of breakthrough infection as data on therapeutic drug monitoring was not captured. Antifungal susceptibility testing was not performed for every cultured isolate, nor were molecular studies (such as CYP51A sequencing to identify mutations) undertaken to characterise azole-resistant isolates or isolates with non-WT MICs for other agents, which would have been useful. The heterogeneity of our study population reduced the capacity to analyze the impact of individual risk factors for mortality. Nonetheless the strength of our study lies in its capture of IA in diverse patient groups and in its contemporary nature across two countries.

Conclusion

In summary, improved diagnostic methods and treatment of IA may have improved outcomes of this important infection. Nevertheless mortality remains substantial in immunocompromised hosts. It is also recognized in patients with mild or no immunosuppression. Further observation/research in evolving risks of IA is needed in this group without classical risk factors. Accurate species identification and susceptibility testing of Aspergillus is essential to inform optimal patient outcomes.

Contributors

ST, SC-CA, KH had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: SC-CA, MS.

Data collection: ST, KH, AP, CHH, AM, TK, OM, NB, BM, JC, RH, KM, PW, AC, CLH, SK.

Adjudication committee: SC-CA, MS, KH, OM, AM, BM, JB, ST.

Analysis and interpretation of data: ST, SC-CA.

Statistical analysis: ST, TS.

Drafting of manuscript and final approval: ST, SC-CA, MY, LW, MS.

Critical revision of manuscript: All authors.

Data sharing statement

Deidentified patient data and data dictionary may be provided upon reasonable request via email to the corresponding author.

Disclosure of individual author:

ST is supported by the University of Melbourne to undertake PhD study, and has received a grant from Gilead. SC has received educational grants from F2G and MSD, Australia, and is the Editor-in-chief of Medical Mycology. AM is in the leadership role of Health Quality and Safety Commission, Southern Cross Healthcare and Australasian Society infectious Diseases NZ branch. OM has received grants/honoraria/consulting fees from Gilead and Merck, Sharp and Dohme Australia. SK is the secretary of the Australia New Zealand Mycoses Interest Group (Australasian Society Infectious Diseases), and has received honoraria/support from Gilead/Pfizer and is on the Data Safety and Monitoring Board (DSMB) for Gilead. TS has received consulting fees from Biogen. PW has received honoraria from ESCMID/ASM. JB has received the National Health and Medical Research Council (NHMRC) Investigator Grant and honoraria from Gilead. MS has received grants and honoraria from F2G, Gilead, Merck, Takeda, and is on the DSMB for Pfizer, Roche and Merck. All other authors declare no competing interests.

Declaration of interests

This work itself is not funded. All authors declare no conflicts of interest associated with this publication, or any financial support that could have influenced its outcome.