COVID-19–Associated Pulmonary Aspergillosis, March–August 2020

1Cases of pneumonia caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) were first described in Wuhan, China, at the end of December 2019. The infection rapidly spread, causing the coronavirus disease (COVID-19) pandemic.

2Because SARS-CoV-2 and treatments such as dexamethasone or tocilizumab can impair the immune system, some researchers anticipated the possibility of fungal superinfection among COVID-19 patients. As of August 2020, researchers have documented COVID-19–associated pulmonary aspergillosis (CAPA), invasive candidiasis, coccidioidomycosis, fusariosis, histoplasmosis, mucormycosis, pneumocystosis, and saccharomycosis. Varying cumulative rates of CAPA have been described, including rates of 0.7%–7.7% among COVID-19 patients, 2.5%–39.1% among ICU patients with COVID-19, and 3.2%–29.6% among COVID-19 patients on mechanical ventilation. Many of these patients lack the concurrent conditions usually associated with invasive pulmonary aspergillosis (IPA) such as malignancies, neutropenia, or history of allogeneic stem cell or solid organ transplantation. Admission to the ICU or severe influenza are also risk factors for IPA in nonneutropenic patients. Reports of CAPA have been mostly limited to a few single-center studies; therefore, a comprehensive analysis of international distribution currently is lacking.

3We analyzed reports in the literature (–; references 51–54in Appendix) and the FungiScope registry (reference 55 in Appendix) to describe baseline conditions, clinical management, and associated deaths in CAPA patients. This analysis also contextualizes the available cumulative incidences.

4We conducted a retrospective analysis using clinical data of patients worldwide who received a CAPA diagnosis during March 1–August 31, 2020. Our analysis comprised data from the FungiScope registry and academic literature (Figure 1).

5FungiScope (https://www.clinicaltrials.gov; National Clinical Trials identifier NCT01731353) is a global registry for emerging invasive fungal infections. FungiScope was approved by the local ethics committee of the University of Cologne, Cologne, Germany (study ID 05-102). The registry includes patients with invasive aspergillosis since 2019. FungiScope’s methods have been described previously (reference 55 in Appendix).

6In addition, we conducted a literature search using the PubMed database (https://pubmed.ncbi.nlm.nih.gov) for suspected CAPA cases occurring in March–August 2020. We used the search string “(Aspergill*) AND (invasive OR putative OR probable OR infection OR case OR patient OR report) AND (COVID* OR corona* OR SARS-CoV-2),” which identified 59 published articles. We reviewed and extracted relevant data from each of the publications. When necessary, we contacted authors for additional details (Appendix).

7We reviewed each patient report using multiple diagnostic definitions. First, we evaluated the patients according to the consensus definition of Koehler et al. (reference 56 in Appendix); we classified the patients as having proven, probable, or possible CAPA. We used alternative definitions to evaluate patients who were nonclassifiable because of lack of essential information, such as the volume of saline recovered by nondirected bronchial lavage (NBL) fluid applied. We categorized the nonclassifiable patients as proven or probable according to the European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group and the National Institute of Allergy and Infectious Diseases Mycoses Study Group criteria for invasive fungal infections or as proven, putative, and colonized according to the AspICU algorithm for IPA in critically ill ICU patients by Blot et al.. We considered severe COVID-19 with acute respiratory distress syndrome (ARDS) to be a valid host criterion (i.e., acquired immunodeficiency). We considered patients who met >1 definition to have CAPA; we categorized the rest as nonclassifiable.

8We collected data on patients’ demographic characteristics and baseline conditions. We also collected data on abnormal radiographic images, mycologic evidence, signs and symptoms at CAPA diagnosis, site of infection, antifungal susceptibility testing, antifungal treatment, death at 6 and 12 weeks after CAPA diagnosis, and absolute death. In addition, we calculated the length of time between COVID-19 and CAPA diagnoses, CAPA diagnosis and most recent healthcare contact with the patient, ICU admission and CAPA diagnosis, and installation of mechanical ventilation and CAPA diagnosis. The contribution of CAPA to patient death (i.e., attributable mortality) was assessed by the treating medical team (Appendix Table 2). To determine the cumulative incidence of CAPA in the facilities included in the analysis, we asked each institution for 3 different denominators: the total number of COVID-19 patients, the number of COVID-19 patients admitted to the ICU, and the number of COVID-19 patients admitted to the ICU who needed mechanical ventilation during March–August 2020.

9We did not calculate an a priori sample size for this exploratory study. To analyze the demographic and clinical characteristics of patients with CAPA, we describe categorical variables using frequencies and percentages; we describe continuous variables using medians and interquartile ranges (IQRs). We used SPSS Statistics 25.0 (IBM, https://www.ibm.com) for statistical analyses.

10We identified 186 CAPA cases during March 1–August 31, 2020, in 17 different countries, according to European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group and the National Institute of Allergy and Infectious Diseases Mycoses Study Group criteria, Blot et al. algorithm, and Koehler et al. consensus definition (reference 56 in Appendix) (Figures 1, 2; Appendix Table 1). We identified 62 (33.3%) cases from literature, 45 (24.2%) from the FungiScope registry, and an additional 79 (42.5%) in both sources (Table 1). The median age among persons with CAPA was 68 years (IQR 59–73 years; range 15–87 years). Most (135; 72.6%) patients were men (Table 2).

11Nearly all (182; 97.8%) patients were admitted to the ICU, most for ARDS (180; 96.8%) or mechanical ventilation (175; 94.1%). Other common baseline conditions and characteristics included corticosteroid administration (98; 52.7%), chronic cardiovascular disease (94; 50.5%), renal failure (74; 39.8%), diabetes mellitus (64; 34.4%), and obesity (47; 25.3%). Overall, 40 (21.5%) patients had chronic pulmonary disease (Table 2).

12In total, 110 (59.1%) patients received either hydroxychloroquine (98; 52.7%) or chloroquine (12; 6.5%) for treatment of COVID-19. Sixty-eight (36.6%) patients received corticosteroids, mainly methylprednisolone monotherapy (26; 14.0%) or antivirals (67; 36.0%), especially lopinavir/ritonavir monotherapy (56; 30.1%). COVID-19 treatment had a median duration of 7 days before recovery or death (IQR 6–11 days; range 1–32 days) (Table 2; Appendix Table 3).

13In 152 (81.7%) patients, CAPA was diagnosed a median of 10 days (IQR 5–16 days; range 0–51 days) after a positive respiratory sample for SARS-CoV-2 infection by reverse transcription PCR. Among all patients, Aspergillus fumigatus was the most frequently reported (122/152; 65.6%) pathogen. Six patients (3.2%) had cultures positive for >1 Aspergillus species. Samples mainly were from bronchoalveolar lavage (BAL) (50; 26.9%), tracheal aspirates (48; 25.8%), or bronchial aspirates (34; 18.3%). In 55 (29.6%) patients, culture was the only diagnostic tool that produced a positive result. Galactomannan (GM) levels were positive (i.e., optical density index ≥1.0) in samples from 113 (60.8%) patients, including BAL samples from 63 (33.9%) patients, serum or plasma from 29 (15.6%), and NBL from 22 (11.8%). Histologic techniques were used for diagnosis in 7 (3.8%) cases. Abnormal radiographic imaging was found in 182 (97.8%) patients, either in computed tomography scans (94; 50.5%), in chest radiographs (48; 25.8%), or both (40; 21.5%) (Table 2).

14Overall, 30 (16.1%) patients provided samples for >1 antifungal susceptibility test, such as microdilution according to European Committee on Antimicrobial Susceptibility Testing guidelines (20; 10.8%) (reference 57 in Appendix), Etest (11; 5.9%), and Clinical and Laboratory Standards Institute microdilution procedures (1; 0.5%) (reference 58 in Appendix). The tests were predominantly performed on A. fumigatus (29; 15.6%) isolates, 3 of which had the TR34L98H resistance mutation in the cyp51A gene. One (0.5%) patient had voriconazole-resistant A. lentulus (MIC 2 µg/mL by EUCAST guidelines) (Appendix Table 4).

15Of 186 CAPA patients, 49 (26.3%) patients did not receive mold-active antifungal therapy. The most common treatments were triazoles (117; 62.9%), especially voriconazole (98; 52.7%, including 79 patients for whom voriconazole was a first-line treatment) and isavuconazole (23; 12.4%). In total, 34 (19.4%) patients received amphotericin B, especially liposomal amphotericin B (23; 12.4%). Of the patients who received amphotericin B, 15 (65.2%) received it as first-line treatment. Antifungal treatment was administered for a median of 16 days before recovery or death (IQR 10–33 days; range 1–92 days) (Table 2; Appendix Table 5).

16In total, 97 (52.2%) patients died, most (89; 47.8%) <6 weeks after CAPA diagnosis. In 32 (17.2%) patients, death was attributed to Aspergillus; including 25 (13.4%) patients who died of aspergillosis and COVID-19 infection. Patients were observed for a median of 22 days (IQR 7–42 days; range 0–144 days) after CAPA diagnosis; survivors were treated for a median of 40 days (IQR 28–50 days; range 1–144 days) and patients who died for a median of 9 days (IQR 3–18 days; range 0–144 days) (Table 2).

17In total, 19 of 39 institutions provided denominators for cumulative incidence over the duration of the study period. The CAPA incidence among all COVID-19 patients ranged from 0.1%–9.7%. Among COVID-19 patients admitted to ICU, cumulative incidences ranged from 1.0%–39.1%. Among patients admitted to ICU who needed mechanical ventilation, cumulative incidences ranged from 1.1%–47.4% (Table 3).

18We described 62 CAPA cases in the literature, 45 in the FungiScope registry, and 79 in both that were diagnosed during March 1–August 31, 2020. Men had a higher (2.6:1) prevalence of CAPA than women. This finding corresponds with a meta-analysis of >3 million COVID-19 patients that showed that men were at increased risk for severe COVID-19 and therefore complications such as CAPA (reference 59 in Appendix).

19Most (97.8%) patients were admitted to the ICU, mainly because of ARDS, need for mechanical ventilation, or both. We found that corticosteroid administration, chronic cardiovascular disease, renal failure, diabetes mellitus, and obesity were common characteristics among these patients. Approximately 1 in 5 patients had chronic pulmonary disease. Patients had many similarities to influenza-associated pulmonary aspergillosis (IAPA) patients from Schauwvlieghe et al., including similar rates of mechanical ventilation (IAPA 90.0% vs. CAPA 94.1%), corticosteroid administration (IAPA 56.0% vs. CAPA 52.7%), baseline renal failure (IAPA 42.0% vs. CAPA 39.8%), obesity (IAPA 30.0% vs. CAPA 25.3%), and chronic pulmonary disease (IAPA 16.0% vs. CAPA 21.5%). IAPA patients had a higher proportion of malignancies (30.0% vs. 11.3%) and solid organ transplantation (13.0% vs. 2.7%); however, CAPA patients had a higher prevalence of diabetes mellitus (12.0% vs. 34.4%). In our study, 50.5% of patients had chronic cardiovascular disease. These differences in the distribution of baseline characteristics between IAPA and CAPA patients reflects the epidemiology of COVID-19, which is more common among those with chronic cardiovascular disease, whereas hematologic or oncologic malignancies are more common among those with IAPA (reference 60 in Appendix). Only 2% of COVID-19 patients have cancer (reference 61 in Appendix).

20Available guidelines for aspergillosis management recommend diagnostic procedures such as respiratory culture and galactomannan index of BAL samples (references 60,62 in Appendix). However, these procedures have a high risk for aerosolization; safety precautions should be used when handling samples from COVID-19 patients (references 63,64 in Appendix). The elevated risk for SARS-CoV-2 transmission and the initial recommendation against using bronchoscopy for COVID-19 diagnosis (references 63,64 in Appendix) might explain the low number of BAL tests used to diagnose CAPA in our study. Schauwvlieghe et al. diagnosed IAPA by using BAL cultures in 63.0% of the patients and the galactomannan test in 88.0%. In the current study, BAL cultures tested positive for Aspergillus in 26.9% of COVID-19 patients; galactomannan tests were positive in 33.9% of patients. Alternative respiratory sample sources (e.g., bronchial aspirate, NBL, tracheal aspirate, and sputum) were used for cultures in 35.4% of IAPA patients and 31.2% of CAPA patients. Alternative samples also were used for galactomannan tests in 17.2% of CAPA patients; if optical density index cutoff values were not standardized for alternative samples, clinicians used the values for BAL. Almost all (97.8%) patients had imaging abnormalities; however, many had only marginally typical features of aspergillosis, hampering the differential diagnosis of CAPA according to radiologic criteria.

21Positive isolates were recovered from 81.7% of CAPA patients. Similar to IAPA patients, the most common (80.3%) pathogen was A. fumigatus. In total, 5 patients had azole-resistant infections: 4 A. fumigatus and 1 A. lentulus infection. We noted 2 patients who had a possible previous exposure to triazoles. The professions of these 2 patients involved exposure to fungicides and manipulated organic matter containing triazole-resistant A. fumigatus. Therefore, the treating teams hypothesized that workplace exposure might have contributed to these patients’ illness. We found a similar proportion of patients with previous azole exposure as Verweij et al. (reference 65 in Appendix); however, the proportion found by Verweij et al. should be considered with caution because of small sample size.

22Triazoles, especially voriconazole, were the most frequently administered antifungal drugs: 52.7% of the study cohort and 71.5% of the patients on antifungal treatment received voriconazole. We found that voriconazole use was associated with decreased death. The first-line use of voriconazole in 79 (80.6%) of 98 patients aligns with current recommendations (references 56,60,62 in Appendix).

23We found a 50% mortality rate at 12 weeks after CAPA diagnosis. This finding is similar to the 51.0% mortality rate of IAPA patients in the same timeframe; however, these rates are almost 20 points higher than in other cohorts, such as aspergillosis patients with acute leukemia (33.8%) (reference 66 in Appendix). Nonetheless, in our study CAPA was attributed as the main reason for death in only 17.2% of the patients, whereas in Koehler et al. (reference 66 in Appendix), it was the main cause of death for 26.9% of patients with hematologic conditions.

24We found an overall 6.9% cumulative incidence for CAPA among patients during the study period, although incidences varied by institution (1.0%–39.1% of CAPA patients admitted to ICU). In most facilities, the rates of CAPA were lower than those of IAPA (14%–19%) (reference 67 in Appendix). However, these ranges might vary according to diagnostic protocols in the different countries and healthcare facilities. Differences in screening practices for CAPA in COVID-19 patients might have affected detection rates and therefore our calculations of cumulative incidence. Further analyses are necessary to establish the geographic variance of this rate.

25The first limitation of this study is that, because of the cross-sectional design of this study, we could not control for disease severity. Second, samples from the lower respiratory tract are the best way to differentiate between colonization and infection, but a low percentage of patients in this study had mycologic evidence from BAL culture or galactomannan tests. Third, we analyzed many cases from literature and could not contact certain authors for further details. In addition, institutions might not have documented all CAPA cases in the literature or FungiScope registry. Given the regional variability of the patient distribution, longitudinal studies might be a more appropriate tool to determine rates. Finally, because of the retrospective nature of the study, we could not retrieve the necessary clinical and diagnostic details of all patients. As a result, many patients were not classifiable according to the definitions used in this article, possibly contributing to an underdiagnosis of CAPA.

26In conclusion, we described a large cohort of CAPA patients using cases from the literature and the FungiScope registry. CAPA occurs mostly in ICU patients on mechanical ventilation. We found that CAPA patients had high rates of chronic cardiovascular disease, renal failure, diabetes mellitus, and corticosteroid use. We also found that CAPA substantially contributed to a high death rate in COVID-19 patients, although cumulative incidence varied by treatment site. We believe that improved screening can identify and enable early treatment of CAPA.