1 Background

COVID-19-associated pulmonary aspergillosis (CAPA) has been reported throughout the COVID-19 pandemic. Based on the multicenere cohort studies, it is estimated that approximately 10% to 15% of patients who develop critical COVID-19 and are treated on intensive care units (ICUs) are diagnosed with CAPA [1-4] and up to 35% of those who receive invasive mechanical ventilation (IMV) [5]. However, incidence rates vary widely by centre, as shown before for the three centres participating in this study [6]. Several characteristics of a SARS-CoV-2 infection are increasing the risk for CAPA development including viral cell tropism, respiratory epithelial damage and the release of pro-inflammatory cytokines as a consequence of release of danger-associated molecular patterns [7, 8]. The outcome is poor with mortality rates greater than 40% and CAPA being an independent risk factor for death in most studies [2, 3, 9, 10].

As invasive procedures for tissue sampling (e.g., lung biopsy) and consecutive histological work-up are rarely feasible in patients with acute respiratory failure (ARF), CAPA diagnosis is usually based on work up of respiratory samples obtained by bronchoscopy including fungal culture, polymerase chain reaction and biomarker testing as well as imaging (although often unspecific) [1]. The majority of CAPA patients present an impeded fungal growth in histopathology from lung tissue and usually lack angio-invasion [11] – the hallmark of invasive aspergillosis in neutropenic patients. Consequently, testing for fungal blood-biomarkers comes along with limited sensitivity [1] and suspicion for CAPA should trigger sampling and testing of lower respiratory tract specimen. Nevertheless, diagnosis remains challenging andrates of CAPA among patients with COVID-19 requiring ICU admission are increasing since the introduction of COVID-19 vaccination programs [12]. A recently published prospective observational study highlighted, that CAPA rates more than doubled (59% vs. 24%) since the population-wide roll out of COVID-19 vaccination programs [13]. This finding is at least partly explained by the fact that patients who have underlying immunocompromising condition are less likely to adequately immunologically response to COVID-19 vaccination [14] and are therefore more likely to develop critical disease requiring treatment on ICU. Consequently, a higher percentage of patients with critical COVID-19 will have underlying conditions that are associated with increased risk for invasive fungal infections and increased mortality.

Given the high burden of CAPA in patients with COVID-19-associated ARF, the high CAPA mortality rate and the challenges in rapid and reliable diagnosis, antifungal prophylaxis was introduced in some centres [15-18]. However, the only randomised controlled trial evaluating the impact of prophylaxis had to be terminated before significant enrolment [19]. Several case reports and case series have reported on inhaled amphotericin B (lipid formulations and conventional) prophylaxis and most found clinically relevant reduction of CAPA rates [15, 16, 18]. One single-centre observational study including non-ventilated patients reported on posaconazole (POSA) prophylaxis in COVID-19 patients on ICU and observed that there was clear reduction of CAPA cases [17].

Here, we report the results of our matched case–control analysis from a prospective multicentre study investigating the potential effect of POSA prophylaxis on the risk of CAPA in mechanically ventilated COVID-19 patients who received corticosteroids.

2 Methods

3 Results

3.1 Study Population

Between September 2020 and April 2021, 83 consecutive cases were identified at centre 1 who received POSA prophylaxis (median 13 days, range: 1 to 48 days) and fulfilled the inclusion criteria. Centre 2 provided 239 potential controls and centre 3 provided data from 192 potential controls (Figure 1). All cases (centre 1) were treated with corticosteroids. The total of 83 patients receiving POSA prophylaxis from centre 1 was matched to 83 patients from centres 2 and 3 each.

Demographic data for the matched study cohort are displayed in Table 1. Of note, the proportion of patients on ECMO differed significantly between centres (16% in centre 1 vs. 2% and 4% in centre 2 and centre 3, respectively, p < 0.001).

TABLE 1. Demographic and clinical characteristics of patients after 1:1 (A) matching and 1:1:1 (B) matching (sensitivity analysis).

Variablesa	Centre 1 (n = 83)	Centre 2 (n = 83)	Centre 3 (n = 83)	p	SD (Centre 1 vs. 2)	SD (Centre 1 vs. 3)
(A)
Posaconazole prophylaxis	83 (100)	0 (0)	0 (0)	< 0.001	Inf.	Inf.
Age at ICU admission in years, median (IQR)b	65 (58–71)	66 (61–71)	65 (57–72)	0.949	0.016	−0.122
Male sexb	56 (68)	54 (65)	55 (66)	0.948	0.043	0.064
EORTC/MSGERC risk factor present at ICU admissionc	8 (10)	5 (6)	5 (6)	0.593	0.148	0.148
ECMO	13 (16)	2 (2)	3 (4)	0.001	0.408	0.504
Treatment with tocilizumabb	2 (2)	1 (1)	3 (4)	0.600	−0.117	0.082
Systemic corticosteroid treatment	83 (100)	83 (100)	83 (100)	—	0	0
Time at risk in days, median (IQR)b	15 (10–23)	15 (10–23)	15 (10–23)	—	0	0

Variablesa	Centre 1 (n = 83)	Centre 2 (n = 83)	Centre 3 (n = 83)	p	SD (cases vs. controls)
(B)
Posaconazole prophylaxis	83 (100)	0 (0)	0 (0)	< 0.001	Inf.
Age at ICU admission in years, median (IQR)b	65 (58–71)	65 (58–71)	65 (58–72)	0.988	−0.041
Male sexb	56 (68)	53 (64)	59 (71)	0.610	0
EORTC/MSGERC risk factor present at ICU admissionc	8 (10)	7 (8)	4 (5)	0.490	0.108
ECMO	13 (16)	2 (2)	4 (5)	0.003	0.408
Treatment with tocilizumabb	2 (2)	1 (1)	2 (2)	0.815	0.042
Systemic corticosteroid treatment	83 (100)	83 (100)	83 (100)	—	0
Time at risk in days, median (IQR)b	15 (10–23)	15 (10–23)	15 (10–23)	—	0

• Abbreviations: ECMO, extracorporeal membrane oxygenation; EORCT, European Organization for Research and Treatment of Cancer; ICU, intensive care unit; IQR, interquartile range; MSGERC, Mycoses Study Group Education and Research Consortium; SD, standardised difference in means or proportions divided by standard error; imbalance defined as absolute value greater than 0.20 (small effect size).
• ^a Expressed as n (%), unless otherwise indicated.
• ^b Matching variables.
• ^c Presence of missing values (0/83 for centre 1, 0/83 for centre 2, 1/83 for centre 3).

4 Discussion

This case–control study investigated the impact of POSA prophylaxis on risk of CAPA development in mechanically ventilated COVID-19 patients receiving corticosteroids. We observed a high inter-centre variability of CAPA incidence rates among the centres; particularly among the two control centres not using prophylaxis. These findings highlight the importance of a nuanced and centre specific management strategy for CAPA.

CAPA is a well-known potential life-threatening complication primarily affecting COVID-19 patients with ARF who require invasive mechanical support [1-3]. As reliable diagnostic methods in non-invasive specimen (e.g., blood) lacks satisfactory clinical and diagnostic performance and imaging patterns are often unspecific, antifungal prophylaxis may contribute to successful management of such patients. Posaconazole is a widely used antifungal agent for mould-active prophylaxis. It has proven efficacy and safety in the haematological malignancy setting [25-27] and is therefore considered standard of care in high-risk patients [28]. Based on its favourable pharmacokinetics, the generally good safety profile and the high concentration of POSA in the cell membrane of leucocytes [29, 30] POSA may also be a favourable agent for CAPA prophylaxis on ICU. Thus, POSA prophylaxis has been established as part of management in COVID-19 ARF patients at centre 1. Prior investigations at centre 1 have shown that CAPA rate dropped locally after implementation of POSA prophylaxis from approximately15%–25% to < 1% [17, 31]. These results indicate that the baseline CAPA rates in centre 1 would have been comparable to that of centre 3 in case there would have been not implementation of POSA prophylaxis. Centre 2, however, turned out to have a low CAPA incidence rate, even lower than the incidence rate in centre 1, despite the fact that centre 2 has not implemented antifungal prophylaxis as standard of care for COVID-19 ARF patients. All study patients received systemic corticosteroid treatment for COVID-19 ARF, which is maybe the most important risk factor [32], but only very few patients in centre 1 (2%), and consequently also among matched controls form centres 2 and 3, received additional tocilizumab treatment. Thus, as anti-IL-6 treatment is a relevant risk factor for CAPA development [2, 33] one may speculate that the study cohort investigated had a low baseline risk for CAPA development. This observation may in fact be underlined by previously reported data from centre 2, where CAPA incidence rate was higher in the sub-cohort receiving tocilizumab versus those not receiving tocilizumab (2.63 versus 1.16 CAPA/1000 ICU days) [6].

In addition, the high variation in CAPA incidence rate among the two control centres may be explained by factors like variation in diagnostic approaches (e.g., biomarker availability [34-36], bronchoscopy strategies [37, 38], etc.) which were not standardised among the centres. Variations in diagnostic strategies may also be a relevant reason why there was no possible CAPA case in centre 2 versus 19 possible CAPA cases in centre 3 before matching. Particularly rate of BALF sampling has been shown to be a very strong driver of CAPA incidence [37], however, in this study rate of BALF sampling were high in all three participating centres, while rate of BALF-GM testing varied between centres. Besides the variations in diagnostic strategies, variations in genetically determined predisposing factors, like aberrations in pathogen-recognition receptors [39], and variations in environmental fungal exposure contribute to the net-state of CAPA risk development in critically ill COVID-19 patients. Besides all the factors mentioned, a standardised diagnostic approach using standardised diagnostic criteria like the recently published FUNDICU consensus definitions are considered an important step for future standardisation and furthermore comparability of study results [40].

In addition, the proportion of patients on ICU with COVID-19 ARF who have an underlying immunosuppressive disease has been increasing during the COVID-19 vaccination era [13]. We therefore investigated the impact of presence of EORTC/MSGERC risk factors on the risk for CAPA development and observed a more than 4-fold increase in odds to develop CAPA for those with EORTC/MSGERC risk factors versus those without. These finding indicate that stratifying antifungal prophylaxis by presence of EORTC/MSGERC risk factors may be a viable option for centres with low CAPA prevalence rates that fall below the threshold to justify universal antifungal prophylaxis for all ventilated patients with COVID-19.

Most of the currently available evidence on antifungal prophylaxis in critically ill COVID-19 patients is showing a reduction of CAPA rates (generally prevalence rates are given) but no significant difference in overall survival rates. While lack of statistical power and small sample sizes may be the most important of this lack of association, there may be other factors that play a role as well. First, most of the studies currently available were performed in centres with a lot of experience and a high awareness for invasive fungal infections. The high awareness and in house available diagnostic tests (e.g., CT-scans, fungal biomarkers, etc.) will usually facilitate a rapid diagnosis, limiting treatment delay and mortality. Autopsy studies in patients with viral-associated pulmonary aspergillosis highlighted, that in the absence of classical risk factors like prolonged neutropenia, Aspergillus spp. usually shows an impeded fungal growth pattern in patients with CAPA [11]. This in some extent complicates early diagnosis as positive serum biomarkers are less likely and further diagnostic procedures are warranted. However, in case of delayed antifungal treatment or missed diagnosis, CAPA is a life-threatening and a deadly disease. Second, the recent studies investigated all COVID-19 patients on ICU, regardless of the required underlying respiratory support. This is a valid approach; however, we have learned that the extent of lung tissue damage and consequently need for respiratory support impacts the risk for the development of CAPA [9]. This study therefore focused on ventilated patients only. Future multicentre studies, optimally randomised controlled trials, are therefore needed to investigate the impact of antifungal prophylaxis on mortality in a wider variation of centres, including centres with less in house availability and longer turnaround times of fungal diagnostics. Finally, the studies are needed for centres with low CAPA prevalence, that focus prophylactic efforts on subpopulations at higher risk, like patients with EORTC/MSGERC host factors.

This study is not exempt from limitations. For example, by employing a case–control design the number of CAPA cases was reduced compared with the numbers originally observed in centre 2 and centre 3. Thus patients included from centre 2 and centre 3 may be not representative of the local epidemiology. However, the case–control design was eventually preferred in order to reliably assess any possible impact of POSA prophylaxis in our cohort. Indeed, this has allowed to avoid a selection bias related to the high baseline proportion of patients treated with tocilizumab (an already recognised risk factor for CAPA) from centre 2 (no prophylaxis), that could have biased results towards a favourable effect of prophylaxis. Of note, without matching, this bias could have not been adjusted/reduced in multivariable analysis due to collinearity between centre (i.e., prophylaxis present versus absent) and tocilizumab use. At the same time, assuming that tocilizumab was given to patients with more severe disease in centre 2, we cannot exclude that our approach resulted in a selection of a lower risk cohort from centre 2 and that the higher CAPA rate in those with tocilizumab was a result of other factors than IL-6 inhibitor treatment. Overall, the potential role of POSA prophylaxis in mechanically ventilated COVID-19 patients treated with tocilizumab may merit a dedicated investigation in this specific subgroup, which was unfeasible in the present study due the very low number of tocilizumab-treated patients receiving POSA prophylaxis. In addition we observed a high variation of ECMO cases among the centres may influencing the baseline risk for CAPA development. It should be noted that we had no influence over the local CAPA screening strategies implemented at each centre. While the number of lower respiratory tract samples sent for mycological testing was comparable across centres, variations in the timing and intensity of screening may have occurred. These differences could have influenced the detection rates of CAPA cases during routine clinical work-up. Lastly, the final analyses reported here differed slightly from the original analyses plan (available at clinicaltrials.gov) based on missing data in the final data set.

5 Conclusions

POSA prophylaxis may be a suitable approach to reduce CAPA rates among COVID-19 patients with ARF depending on the local characteristics of COVID-19 patients on ICU and the local epidemiology. Further investigational trials are needed that investigate the efficacy of POSA prophylaxis in pre-selected patients with certain baseline characteristics that come with increased risk for the development of invasive fungal infections.

Author Contributions

Juergen Prattes: conceptualization, methodology, writing – original draft, funding acquisition, project administration. Daniele R. Giacobbe: methodology, formal analysis, writing – original draft, project administration. Cristina Marelli: formal analysis. Alessio Signori: formal analysis. Silvia Dettori: data curation. Greta Cattardico: data curation. Stefan Hatzl: data curation. Alexander C. Reisinger: data curation. Philipp Eller: data curation. Robert Krause: data curation. Florian Reizine: data curation. Matteo Bassetti: methodology, writing – review and editing, supervision. Jean-Pierre Gangneux: methodology, writing – review and editing, supervision. Martin Hoenigl: conceptualization, methodology, writing – review and editing, funding acquisition, project administration, supervision.