Invasive fungal infections among critically ill adult COVID-19 patients: First experiences from the national centre in Hungary

Introduction Data suggests that invasive fungal infections (IFI) might complicate COVID-19. Our goal was to describe characteristics of IFI among critically ill COVID-19 adults. Methods A retrospective observational case-series analysis was done between March–July 2020. Consecutive patients with critical COVID-19 were eligible, and have been included when proven or putative/probable IFI could be confirmed during their course. For COVID-19 diagnosis, ECDC definitions and WHO severity criteria were followed. Candidaemia was diagnosed according to the ESCMID 2012 guideline. Invasive pulmonary aspergillosis (IPA) was defined following EORTC/MSG, ECMM/ISHAM and modified AspICU criteria. Outcome variables were rates of IFIs, in-hospital all-cause mortality, rate and time to negative respiratory SARS-CoV-2 PCR. Results From 90 eligible patients, 20 (22.2%) fulfilled criteria for IFI. Incidence rate for IFI was 2.02 per 100 patient-days at ICU. Patients were mostly elderly males with significant comorbidities, requiring mechanical ventilation because of ARDS. IFI could be classified as candidaemia in 7/20 (40%), putative/probable IPA in 16/20 (80.0%). Isolated species of candidaemia episodes were Candida albicans (4/9, 44.4%), Candida glabrata (3/9, 33.3%), Candida parapsilosis (1/9, 11.1%), Candida metapsilosis (1/9, 11.1%). Mold isolates from lower respiratory tract were Aspergillus fumigatus, BAL galactomannan positivity was prevalent (16/20, 80.0%). Mortality was 12/20 (60.0%) with a median time to death of 31.0±37.0 (5–89) days. Only 9/20 (45.0%) patients reached SARS-CoV-2 PCR negativity after a median time of 20.0±12.0 (3–38) days. Conclusion In this small cohort of critically ill COVID-19 adults, morbidity and mortality related to invasive fungal infections proved to be significant.


Introduction
Coronavirus disease-19 (COVID-19) is a potentially life threatening infection caused by the highly virulent human coronavirus, SARS-CoV-2 (severe acute respiratory syndrome coronavirus-2). In its most severe forms, COVID-19 progresses to multi-systemic dysfunction, including acute respiratory distress syndrome (ARDS) and cytokine storm (CS). Data from the literature suggest that invasive fungal infections (IFI) might also be accounting for additional morbidity in critical COVID-19, especially in patients with the aforementioned complications [1]. Our aim was to assess the burden and characteristics of invasive fungal infections among critically ill adult COVID-19 patients hospitalized at our centre during the first 4 months of the pandemic.

Data collection
A database has been established for the study purpose by manual data extraction from hospital records, and anonymized transfer to a standardized case report form. Collected data were: 1) age and gender, 2) comorbidities, 3) length of stay (LOS), 4) clinical and radiological parameters at COVID-19 and IFI diagnosis (symptom onset, oxygen demand, Horowitz index, ARDS and CS, radiomorphology on chest X-ray or computed tomography [CT]), 5) laboratory parameters at COVID-19 and IFI diagnosis (blood leucocyte, absolute neutrophil granulocyte, lymphocyte and platelet counts, hemoglobin, CRP, procalcitonin, serum ferritin, high sensitivity troponin-I, serum interleukin-6, NT-proBNP, serum creatinine, LDH and D-dimer), 6) rate and time to negative respiratory SARS-CoV-2 polymerase chain reaction (PCR), 7) microbiological characteristics ( [2]. A suspected COVID-19 case is confirmed if SARS-CoV-2 nucleic acid is detected by PCR in a clinical specimen. COVID-19 severity is stratified per World Health Organization (WHO) criteria [3]. Non-infectious complications of critical COVID-19 disease were defined as ARDS and/or CS. ARDS is identified according to the 2012 Berlin criteria, CS is diagnosed by trends of clinical and laboratory parameters along with the HScore [4,5]. IFI was defined if presence of a yeast or mold was proven by culture or non-culture based microbiological methods in a clinical sample obtained by a sterile procedure from a physiologically sterile site/ fluid, along with a compatible clinical presentation. Candidaemia episodes were diagnosed if Candida sp. was recovered from ≥1 blood culture [6]. Cases of invasive pulmonary aspergillosis (IPA) are ascertained following the EORTC/MSG criteria in immunocompromised (proven or probable diagnosis) and modified AspICU criteria in immunocompetent patients (proven or putative diagnosis) [7−9]. In both systems, proven IPA is defined by microscopic analysis (histopathologic, cytopathologic or direct microscopic examination showing hyphae with tissue damage) and/or Aspergillus sp. culture recovery from sterile material obtained by bronchoscopic needle aspiration or lung biopsy. For probable IPA diagnosis, EORCT/MSG criteria rely on host factors (see below) with clinical features (typical patterns on chest CT) and mycological evidence (direct testing: microscopy showing fungal elements or Aspergillus sp. culture recovery from BAL, bronchial brush or tracheal aspirate; indirect testing: galactomannan antigen or Aspergillus sp. PCR from blood, BAL, bronchial brush or tracheal aspirate) of IFI. EORCT/MSG criteria are applied in cases of inherited or acquired severe immunosuppression, such as prolonged neutropenia (eg. active haematological malignancy, allogeneic stem cell transplantation), systemic corticosteroid treatment or T-cell immunosuppressants. Contrarily, other groups of critically ill patients cannot be classified in the absence of host factors. Modified AspICU criteria applicable for this cohort is a clinical algorithm which relies on microbiological evidence of Aspergillus sp. from the lower respiratory tract as entry criterion (original AspICU criteria needed positive culture result from tissue or BAL, modified AspICU criteria added serum/BAL galactomannan antigen positivity), and considers clinical (eg. refractory or recrudescent fever, worsening respiratory insufficiency despite antibiotic therapy etc.) and imaging (pulmonary infiltrates by chest X-ray or CT) criteria for putative IPA (facultative host risk factors were also listed in the original AspICU criteria, similar to EORCT/MSG). After study completion, the 2020 ECMM/ISHAM consensus criteria for COVID-19-associated pulmonary aspergillosis have been published. Therefore, cases have retrospectively been re-evaluated according to these criteria as well [10]. Upper or lower respiratory tract colonizations (cases not corresponding to invasive disease) with either Aspergillus sp. or Candida sp. were not considered for inclusion, per consensus criteria.
At our centre, patients with critical COVID-19 disease are transferred to the ICU from the isolation ward, where on-demand realtime infectious disease consultation is provided to intensivists. To facilitate in-house diagnostic and therapeutic strategies, care of COVID-19 patients was standardised by written protocols and checklists, according to best available literature evidence. From every patient at ICU admission and/or during intubation, 2 sets of blood cultures, a mini-BAL and subsequent BAL for bacterial and fungal cultures and GM, and blood for BDG and GM were obtained on the same day. Mini-BAL is a non-bronchoscopic, blinded lavage technique during which 10-20 mL of saline is inoculated and re-aspirated through the closed breathing circuit of an intubated patient. BAL sampling, along with needle aspiration (if necessary) was performed for infection validation by a pulmonologist expert on a case-by-case basis, if contraindications were not found for bronchoscopy. All patients included in the study received mini-BAL at admission/intubation, and subsequent BAL for verification of pulmonary infection. Serum BDG and GM were tested at admission, and twice weekly thereafter. Positive galactomannan results are confirmed by another clinical sample taken during a separate procedure. Blood cultures, respiratory samples and fungal biomarkers were retaken simultaneously if the patient developed sepsis, had persistent fevers or deteriorated clinically, without other plausible explanations. Sepsis was defined according to SEPSIS-3 criteria, VAP was diagnosed according to the Infectious Disease Society of America (IDSA) guidelines [11,12]. Cultures and non-culture based microbiological diagnostics were executed at the local microbiology laboratory of our centre. Fungal isolation was done on Candida chromagar (CHROMID Candida, bio-M erieux, Spain) and Sabouraud agar (Sabouraud Gentamicin Chloramphenicol 2, bioM erieux, Spain). Fungal identification was done by observation of agar culture characteristics, light microscopic morphology and matrix-assisted laser desorption/ionization time-offlight mass spectrometry (Vitek-MS V3, bioM erieux, Spain). Yeast and mold susceptibilities for amphotericin-B, azoles (excluding isavuconazole) and echinocandins were tested by broth microdilution (MICRONAUT-AM Antifungal Agents MIC, MERLIN Diagnostika, Germany), isavuconazole susceptibility was tested by E-test (Liofilchem, Italy). Fungal susceptibility interpretations were done following European Committee on Antimicrobial Susceptibility Testing (EUCAST) recommendations (www.eucast.org). Beta-D-glucan levels were tested by kinetic turbidimetry, a cut-off value of ≥11 pg/mL corresponded for positivity (b-Glucan Test, FUJIFILM Wako Pure Chemical Corporation, Japan). Galactomannan antigen testing was done by enzymelinked immunosorbent assay, an optical density index of ≥0.5 was interpreted as positivity (Platelia Aspergillus EIA, Bio-Rad, France).

Study outcomes and analysis
Clinical outcome variables were rate and time to in-hospital allcause mortality, and rates of invasive fungal infections. Virological outcomes were rate and time to negative SARS-CoV-2 respiratory PCR sampling. PCR negativity was defined if ≥2 consecutive respiratory PCR samples taken ≥48 hours were proven to be negative. Incidence regarding IFI was calculated by using cumulative data from hospital records. Continuous variables were expressed as median § interquartile range with minimum-maximum limits, categorical variables were expressed as absolute numbers (n) and percentages (%). Given the expected low incidence of IFIs, modelling of the primary outcome was not planned a priori. For reporting, we adhered to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) Statement [13].

Demographic and clinical characteristics
From 90 patients hospitalized with critical COVID-19 at our centre during the study period, 20 (22.2%) patients with a diagnosis of any IFI were identified and included in the study. Incidence was calculated as 2.02 for any IFI episode, 0.91 for candidaemia episodes, and 1.61 for IPA episodes, each per 100 ICU patient-days. Demographic and clinical characteristics are shown in Table 1., laboratory and microbiological characteristics are reported in Table 2

Present study
In this study, conducted among critically ill adult COVID-19 patients hospitalized at a single ICU during a 4-month period, we have calculated that 22.2% of all patients were diagnosed with IFI. Most patients were elderly males, with a relevant comorbidity burden of mostly chronic cardiopulmonary and metabolic diseases. Almost all patients had ARDS and ongoing cytokine storm at diagnosis, prompting for complex intensive care maneuvers and immunomodulatory therapies. Among 20 enrolled cases from 90 eligible patients, 35.0% had at least on episode of candidaemia with 4 distinct Candida species. Serum BDG positivity was 55.0%. Probable/putative IPA was diagnosed in 80.0% among 20 included patients. Three patients had BAL fungal cultures positive for Aspergillus fumigatus. Serum galactomannan antigen testing had a lower diagnostic yield for IPA (40.0%), while respiratory galactomannan antigen testing was more sensitive (80.0%). Some patients had invasive bacterial coinfections, further complicating adequate antimicrobial strategies. In our cohort, all-cause mortality was high (60.0%) at 31.0 §37.0 days.
Previous studies from the literature According to a review by Antinori et al. examining 33 cases with COVID-19 associated IPA from the literature, all patients had been mechanically ventilated at an ICU ward. Patients were predominantly elderly males (81%) with a history of chronic obstructive pulmonary disease (21%) and diabetes mellitus (27%) as comorbidities. The Table 2 Laboratory and and microbiological characteristics of adult patients with critical COVID-19 and invasive fungal infections included in the study.

Parameters
Total (      authors calculated an overall mortality of 67% among those affected, with Aspergillus fumigatus isolated from BAL or tracheal samples as the causative pathogen of IPA in most cases. Serum galactomannan positivity was low (23%), while respiratory galactomannan testing had a better yield (71%). Patients were diagnosed by using the AspICU algorithm [14]. In our cohort, we found a similar rate from serum, but tracheal sampling provided lower positivity compared to results of Antinori et al. Therefore, we cannot exclude a higher rate of false negative testing at our centre (eg. due to mucolytic agent use among patients not included in the final cohort). Incidence and outcomes of IPA among critically ill adult patients with COVID-19 widely vary between centres. In a retrospective chart review done by Koehler et al., putative IPA was diagnosed by AspICU criteria in five of 19 (26.3%) consecutive cases among patients with COVID-19 associated ARDS admitted to the ICU. Voriconazole, isavuconazole and caspofungin were initiated as antifungal therapies. Three patients died. All identified isolates were Aspergillus fumigatus [15]. In contrast, van Arkel et al. observed an incidence of 19.4% with 3 probable and 3 possible IPA cases using the EORTC/MSG criteria, among a cohort of 31 ICU patients. Aspergillus fumigatus could be identified by culturing in five cases as the causative pathogen, mostly voriconazole with anidulafungin was administered. Four patients died (66.7%) [16]. In our cohort, similarly to literature results, all mold isolates recovered from BAL samples were Aspergillus fumigatus. Empirical antifungal therapy was mostly amphotericin-B, but targeted deescalation to mostly voriconazole could be performed in some cases. A larger cohort of patients was analysed by Wang et al. They found that among 104 patients with COVID-19, 8 (7.7%) had IPA, with risk factors of older age, initial b-lactam/lactamase inhibitor combination therapy, mechanical ventilation and COPD [17].
Data concerning incidence and outcome of candidaemias among critically ill COVID-19 adult patients are scarce. In a review by Lai et al. detailing 14 published studies, candidaemia was detected among 4.0% of mostly ICU patients, while a retrospective case-series of 836 hospitalised COVID-19 patients from two UK hospitals found low numbers of candidaemia: only 3 line-releated infections were documented [18,19]. Agrifoglio et al. documented 15 (10.8%) candidaemia cases with C. albicans, C. parapsilosis and C. glabrata among 139 critically ill patients. Patients were mechanically ventilated, required vasopressors, had implanted central venous catheters and were receiving total parenteral nutrition and corticosteroids for ARDS at candidaemia diagnosis. The overall calculated mortality was 40% [20]. Moreover, Antinori et al. observed a relatively high rate of candidemia (6.9%) among a cohort of 43 COVID-19 patients treated with tocilizumab at the ward or ICU, suggesting that IL-6 blockade might be associated with invasive yeast infections [21]. In our cohort, the rate of candidaemia was higher compared to literature data, possibly reflecting a broader risk factor burden and longer ICU hospitalization times. The choice of empirical antifungal was dominantly caspofungin, targeted fluconazole could be given in some cases.

Limitations
Our study has several limitations. Firstly, this was a case-series analysis done in a single centre. The relatively small sample size and the lack of a matching control group limits exact risk estimation of IFI among COVID-19 patients. Some patients had multiple infectious complications, and a definite cause of death could not be determined, as autopsies of COVID-19 patients were not routinely done during the first wave in Hungary. Some diagnostic uncertainties might have biased our results, eg. the higher rate of serum BDG might partially represent false positivity (eg. due to intravenous immunoglobuline, parenteral nutrition, gauze usage etc.). Exact body weights were not routinely documented at ICU, but presence of obesity based on clinical examination was archived. Finally, it could be challenging to distinguish a colonization from a clinical infection with Aspergillus sp. in patients with COPD. Despite these limitations, we think that our study might underscore the importance of clinically relevant, potentially fatal invasive fungal diseases among COVID-19 patients.

Conclusion
In this series of critically ill adult COVID-19 patients, invasive fungal infections including invasive pulmonary aspergillosis and candidaemia seemed to result in relevant morbidity and mortality burden. Further prospective data should be collected on this emerging subject.
Transparency declaration