Invasive Fungal Infections Complicating COVID-19: A Narrative Review

Invasive fungal infections (IFIs) can complicate the clinical course of COVID-19 and are associated with a significant increase in mortality, especially in critically ill patients admitted to an intensive care unit (ICU). This narrative review concerns 4099 cases of IFIs in 58,784 COVID-19 patients involved in 168 studies. COVID-19-associated invasive pulmonary aspergillosis (CAPA) is a diagnostic challenge because its non-specific clinical/imaging features and the fact that the proposed clinically diagnostic algorithms do not really apply to COVID-19 patients. Forty-seven observational studies and 41 case reports have described a total of 478 CAPA cases that were mainly diagnosed on the basis of cultured respiratory specimens and/or biomarkers/molecular biology, usually without histopathological confirmation. Candidemia is a widely described secondary infection in critically ill patients undergoing prolonged hospitalisation, and the case reports and observational studies of 401 cases indicate high crude mortality rates of 56.1% and 74.8%, respectively. COVID-19 patients are often characterised by the presence of known risk factors for candidemia such as in-dwelling vascular catheters, mechanical ventilation, and broad-spectrum antibiotics. We also describe 3185 cases of mucormycosis (including 1549 cases of rhino-orbital mucormycosis (48.6%)), for which the main risk factor is a history of poorly controlled diabetes mellitus (>76%). Its diagnosis involves a histopathological examination of tissue biopsies, and its treatment requires anti-fungal therapy combined with aggressive surgical resection/debridement, but crude mortality rates are again high: 50.8% in case reports and 16% in observational studies. The presence of other secondary IFIs usually diagnosed in severely immunocompromised patients show that SARS-CoV-2 is capable of stunning the host immune system: 20 cases of Pneumocystis jirovecii pneumonia, 5 cases of cryptococcosis, 4 cases of histoplasmosis, 1 case of coccidioides infection, 1 case of pulmonary infection due to Fusarium spp., and 1 case of pulmonary infection due to Scedosporium.


Introduction
SARS-CoV-2, the causative agent of coronavirus disease 2019 (COVID- 19), is responsible for a respiratory disease whose broad spectrum of severity ranges from asymptomatic or mildly symptomatic infection to severe bilateral pneumonia leading to progressive respiratory failure requiring non-invasive or invasive mechanical ventilation [1]. The death toll of COVID-19 is mainly due to extensive lung damage, although a number of studies have suggested that secondary pulmonary or non-pulmonary, bacterial, or fungal infections play a significant role [2,3]. Many factors may be associated with these secondary infections, including immune system dysregulation and inhibition, epithelial barrier damage, the widespread use of antibiotics, admission to an intensive care unit (ICU), mechanical ventilation, and prolonged hospitalisation [1,2,4]. The main etiological agents of secondary infections in COVID-19 are Gram-positive and Gram-negative bacteria, but some yeasts and moulds have also been described as deadly secondary pathogens [5].
Invasive pulmonary aspergillosis (IPA) is a life-threatening disease caused by the ubiquitous Aspergillus mould, which is typically associated with immune system dysregulation due to chronic glucocorticoid therapy, prolonged neutropenia in hematological patients, stem cell/solid organ transplantation, or immunosuppressive treatment [6]. Since the first reports of COVID-19 in Wuhan, some authors have highlighted the risk of invasive fungal infections (IFIs) in critically ill COVID-19 patients on the basis of the increasing number of reports of IPA in ICU patients without immunological disorders, such as those with severe influenza or chronic obstructive pulmonary disease [7][8][9]. Various published case reports and observational studies of COVID-19-associated IPA (CAPA) have raised concerns about the burden of this infection and difficulties in diagnosing it (non-specific clinical and radiological findings, the uncertain diagnostic performance of microbiological assays in an ICU, difficulties in differentiating colonisation from infections, the fact that the current clinical diagnostic algorithm does not really apply to COVID-19 patients, and concerns about collecting lower respiratory tract samples because of the risk of contagion) [10,11].
Candidemia is a typical secondary infection of critically ill patients that is associated with prolonged mechanical ventilation, total parenteral nutrition, broad spectrum antibiotic treatment, in-dwelling blood catheters, glucocorticoid therapy, and abdominal surgery before ICU admission [12,13]. COVID-19 patients admitted to an ICU generally undergo prolonged mechanical ventilation due to severe acute respiratory distress syndrome (ARDS) and receive antimicrobials, and are therefore at greater risk of developing secondary candidemia, which has been reported in a number of studies of COVID-19 [14,15].
Mucormycosis is an invasive fungal infection (IFI) that is caused by caused by naturally ubiquitous filamentous Mucorales fungi. The factors predisposing for mucormycosis include uncontrolled diabetes mellitus, glucocorticoid treatment, malnutrition, and other immunosuppressive conditions [16,17], and an increasing number of reports (mainly from India) have indicated that mucormycosis can be detrimental in patients with ongoing or previous COVID-19, especially if they are diabetic or have poorly controlled glycemia [18].
Finally, Pneumocystis jirovecii pneumonia (PCP), cryptococcosis, and histoplasmosis are other IFIs that have been reported in COVID-19 patients, which suggests that SARS-CoV-2induced immune dysregulation can increase the likelihood of developing opportunistic infections typically seen in patients with severe immunodepression, such as acquired immunodeficiency syndrome (AIDS) or haematological malignancies [19].
The aim of this narrative review is to describe the characteristics of the IFIs complicating COVID-19 by separately analysing case reports and observational studies, with particular emphasis on the clinical and microbiological characteristics of CAPA and the clinical algorithms used to diagnose it.

Materials and Methods
The PubMed and Scopus databases were searched to find English, Italian, or Spanish case reports, case series, or retrospective or prospective observational studies of IFIs occurring in COVID-19 patients that were published between 1 January 2020 and 18 June 2021 using the following terms: (Aspergill* OR CAPA) AND (invasive OR putative OR probable OR infection OR case OR patient OR report) AND (COVID* OR corona* OR SARS-CoV-2); (Candida OR candidemia OR candidiasis) AND (infection OR case OR patient OR report) AND (COVID* OR corona* OR SARS-CoV-2); (mucormyc* OR rhizopus) AND (invasive OR putative OR probable OR infection OR case OR patient OR report) AND (COVID* OR corona* OR SARS-CoV-2); (pneumocyst*) AND (infection OR case OR patient OR report) AND (COVID* OR corona* OR SARS-CoV-2); (cryptococc*) AND (infection OR case OR patient OR report) AND (COVID* OR corona* OR SARS-CoV-2); (histoplasm*) AND (infection OR case OR patient OR report) AND (COVID* OR corona* OR SARS-CoV-2). Adults diagnosed with IFIs and COVID-19 were included, regardless of their age or origin, and, when available, their patient-level demographic, clinical, and microbiological data were extracted. The different IFIs are discussed separately, differentiating case reports and case series from retrospective and prospective observational studies. We also discuss the characteristics of CAPA and the applicability of its various clinically diagnostic algorithms.
Ninety-two percent of the patients received anti-fungal treatment, and there were two pan-azole-resistant infections [75,76]. In the case reported by Mohammed et al. [75], the resistance of the A. fumigatus strain was determined by means of phenotype testing using a 4-well triazole resistance screen (VIP check™) and the minimum inhibitory concentrations (MIC) using gradient strips. Ghelfenstein-Ferreira et al. described an A. fumigatus strain with high MICs for all azoles and identified the TR34/L98H mutation [76].

Observational Studies
Four hundred and four cases of CAPA were described in 47 observational studies of 12,080 COVID-19 patients (average incidence 3.3%); the incidence of CAPA in the individual studies varied from 0.1% [28] to 57.1% [34] ( Table 2; detailed information about the individual studies is given in Table S2). Most of the studies (85%) exclusively involved critically ill patients requiring ICU admission. Given the lack of consensus concerning the criteria for defining CAPA, it was diagnosed using various case definitions and clinical algorithms, including the AspICU algorithm with or without modifications (14 studies) [103], the modified AspICU criteria (7 studies) [104], the influenza-associated IPA (IAPA) criteria of Schauwvlieghe et al. (8 studies) [105], the new CAPA criteria of White et al. (1 study) [30], the algorithm proposed by ECCM/ISHAM (6 studies) [106], and the European Organization for Research and Treatment of Cancer and the Mycoses Study Group Education and Research Consortium (EORTC/MSGE) algorithm (7 studies) [107]. Ten studies did not specify the criteria used to diagnose CAPA [21,27,29,31,36,44,51,57,59,63].
Histopathological or cytological findings were considered in eight studies involving 16 patients: fungal hyphae were detected in the respiratory specimens of 10 (nine BAL, one TA) [33,35,43,47,52,55], trans-bronchial pulmonary biopsy led to a diagnosis of "proven CAPA" in four [24], and the diagnosis was confirmed by means of autopsy in two [20].
As in the case reports and CAPA patient series, chest CT findings suggested IPA was rarely used in the observational studies (18.8% of cases): the main findings were nodules (69%), cavities (38.1%), and wedge-shaped consolidations (16.7%).
Anti-fungal treatment was received by 71.7% of the patients (voriconazole in 64.3% of cases). It is worth noting that two studies reported three cases of azole-resistant Aspergillus-Machado et al. described a voriconazole-resistant isolate of A. lentulus (MIC 2 mg/L) [39], and Meijer et al. described two azole-resistant isolates of A. fumigatus that were positive to the VIPCheck test and molecular biology for the TR34/L98H mutation [49].
The reported crude mortality rate in the observational studies was 54.6% (184/337).

CAPA Classification Criteria
Three hundred and ninety-five CAPA cases were identified on the basis of the main classification criteria used by the authors of each study, and Table S3 summarises the classification criteria that we considered when we reclassified the CAPA cases using the patient-level data available in 35 studies (Table S2) on the basis of the clinical algorithms shown in Table 3. The newly proposed ECMM/ISHAM definition criteria identified a higher number of CAPA cases (236 probable, 79 possible) [106]; the IAPA classification of Verweij et al. [104] identified 211 cases, and that of Schauwvlieghe et al. [105] identified 206 cases. The new CAPA criteria of White et al. [30] identified 180 putative cases of CAPA, whereas the AspICU algorithm [103] identified only 49 putative cases but 163 patients with Aspergillus colonisation (it is the only clinical algorithm incorporating a "colonisation" category). The diagnostic criteria proposed by the EORTC/MSGE [107], which were originally designed and validated in the context of haematological patients, identified only 25 cases (Table 4).
In 18 episodes, Candida albicans was isolated (40.9%), and 24 were due to other Candida species. In three patients from two studies, two Candida species were reported [108,119]. It is also worth noting that two patients were infected by Saccharomyces cerevisiae following treatment with a probiotic mixture for diarrhoea [118]. There were 11 cases of candidemia due to C. auris: two strains in the study of Allaw et al. and three in the study of De Almeida et al. were resistant to fluconazole and amphotericin B [112,114]; all of the six strains reported by Villanueva were resistant to amphotericin B and three were resistant to fluconazole [119]. A secondary infectious focus was described in four patients (one had endophthalmitis, one retinitis, one endocarditis, and one had fungal costochondritis and spondylitis confirmed by bone biopsy histology and culture) [109,113,123]. Anti-fungal treatment was received by 87.8% of the patients, mainly echinocandins (86%).

Case Reports and Case Series
Twenty-four case reports and nine case series [84,88, reported 59 cases of mucormycosis in patients with a median age of 51 years (IQR 42-62), 71.2% of whom were males (Table 6; detailed information about the individual studies is given in Table S6). The most frequent co-morbidity was diabetes mellitus (77.9%).
Histopathological examinations (HPEs) were carried out in all cases, 91.5% of which were positive; the diagnosis of the remaining five cases was supported by cultures of infected specimens (BAL, sputum, and purulent orbital material) [147,165]. Twenty-nine diagnoses of fungal rhinosinusitis were made on the basis of HPEs of nasal biopsies obtained during functional endoscopic sinus surgery (FESS); a further 10 diagnoses were based on HPEs of nasal material obtained by other means (discharges, scrapings, aspirates, or swabs). Definite diagnoses of pleuro-pulmonary mucormycosis were made after HPEs of BAL (three cases), a pulmonary biopsy (one case), pleural effusion (one case), a pleural biopsy (one case), induced sputum (one case), and BAL plus a pulmonary biopsy (one case) [84,88,138,139,141,145,147,149,167].
Cultures were positive in 33/37 patients (89.2%). The most frequent finding was Rhizopus spp. (26 isolates), with the strain being identified in 13 cases: R. microsporus (five cases), R. arrhizus (seven cases), and R. azygosporus (three cases). The species was identified using molecular biology techniques on tissue biopsies (positive PCR) in two cases [138,144]. Anti-fungal treatment was administered to 93.1% of the patients, and 59.3% required surgical interventions (31 maxillo-facial debridements, two thoracic soft tissue debridements, one pleural decortication, and one pulmonary resection).

Observational Studies
The literature search found 12 observational studies, in which 3126 cases of mucormycosis were described [18,45,61,[169][170][171][172][173][174][175][176][177], the average incidence of which was 8.6%. As shown in Table 7 (detailed information about individual studies is given in Table S7), most of the patients were males (70.6%), had diabetes mellitus (76.1%), received systemic steroids for COVID-19 (73.8%), and developed mucormycosis concomitantly with COVID-19. In the largest Indian study, the median time from the diagnosis of COVID-19 to the diagnosis of mucormycosis was 13 days [177]. Rhino-orbital and rhino-sinusal mucormycosis were the most frequent (respectively, 48.9% and 24.6%); pulmonary involvement was reported in only 20 cases (0.6%). The diagnosis was confirmed by means of histopathology in 1340/3062 cases (43.8%), whereas cultures revealed zygomycetes in only 23.6%. The species was identified in 29 cases (5%) and not reported 508 cases. In one case, the diagnosis was made on the basis of a positive Rhizopus/Mucor PCR in a respiratory sample and serum [45]. More than two-thirds of the patients received anti-fungal treatment (75.6%), mainly amphotericin B (97.3%). A surgical intervention was undertaken in 2352/3113 patients (68.3%). All but two of the studies reported mortality data: the overall crude mortality rate was 16%.  [176,177] report the median time from the diagnosis of COVID-19 to the diagnosis of COVID-associated mucormycosisbut not the frequency of patients with cuncurrent or post-COVID-19 infection.
Twenty cases of Pneumocystis jirovecii pneumonia (PCP) were reported, all of them diagnosed during the course of COVID-19; 30% of the patients were HIV-positive at the time of both diagnoses. More than half of the patients required mechanical ventilation. The diagnosis of PCP was established mainly on the basis of P. jirovecii PCR in respiratory samples (80%); histopathological evidence was rare (one autoptic diagnosis, one direct examination of sputum) [183,187]. More than half of the patients received specific treatment (65%), and eight deaths were reported [178][179][180][181][182][183][184][185][186][187][188].

Discussion
As expected, a number of cases of secondary bacterial and fungal infections have been reported since the emergence of COVID-19 and the publication of the first observational studies [90,130,197]; however, to the best of our knowledge, this is the only descriptive review of all of the types of IFI associated with COVID-19, the fungi involved, their clinical presentation and diagnostic difficulties, and their crude mortality rates. We identified 4099 cases of IFIs in 58,784 COVID-19 patients reported in 168 studies. CAPA and yeast fungemia were mainly reported among critically ill COVID-19 patients requiring ICU admission (in respectively 88% and 65% of the studies), whereas only 47% of the mucormycosis studies were carried out in an ICU.
Although it is currently unknown whether SARS-CoV-2 is an independent risk factor, patients with severe COVID-19 seem to be at risk of developing IPA. We identified a total of 478 cases of CAPA in 47 observational studies and 41 case reports/case series. Its average incidence was 3.9%, but its incidence in the observational studies ranged from 0.1% [28] to 57% [34], a variability that is probably due to differences in the diagnostic strategies and clinical algorithms used by the authors.
It is difficult to diagnose IPA in ICU patients as all of the proposed diagnostic algorithms have major limitations (Table S3). The EORTC/MSG criteria do not really apply to COVID-19 patients because they often do not have any host risk factors and the CT criteria are infrequently observed in mechanically ventilated patients [107]. Among the patients we analysed, the EORTC/MSG criteria identified only 25 cases of CAPA (16 probable and 9 possible: 6.3%). In a bid to overcome such limitations, Blot et al. validated the AspICU algorithm, which distinguishes putative IPA from Aspergillus colonisation (the other algorithms do not contemplate a "colonisation" category); however, although its diagnostic performance is better than that of the EORTC/MSG criteria, its main limitation is that it does not include indirect biomarkers of fungal infection (GM or BDG) [103], and it identified only 49 putative cases of CAPA (12.4%).
The two case definitions of influenza-associated pulmonary aspergillosis (IAPA) in ICU patients proposed by Verweij and Schauwvlieghe allowed for a probable CAPA diagnosis in COVID-19 patients admitted to an ICU with pulmonary infiltrates and a positive BAL culture or BAL or serum GM [104,105]. These criteria are less stringent than the EORTC/MSG and AspICU criteria, but they may lead to an over-estimate of the incidence of CAPA and are less specific as patients with just a single positive GM test and chest X-ray infiltrates (frequent findings in critically ill COVID-19 patients) may be classified as having CAPA: applying these two criteria to all of the CAPA studies led to the identification of, respectively, 211 (53.4%) and 206 cases (52.2%), the second highest numbers of all of the classification criteria. Moreover, neither case definition includes the PCR testing of serum or respiratory sample or BDG, whereas it has been shown (at least in hematological patients) that the diagnostic performance of combining multiple biomarkers is better than using just one [198].
White et al. have proposed a new CAPA definition based on radiological and mycological criteria, which includes Aspergillus PCR testing and serum BDG, which in one study of ICU patients has shown to be more sensitive than serum GM in diagnosing aspergillosis [199]. However, although one biomarker may be diagnostically sufficient in the presence of a specific radiological feature, this new CAPA definition requires at least two positive mycology results to be sufficiently specific in the presence of non-specific chest radiology findings [30]. Nevertheless, in comparison with AspICU, it identified 131 more cases.
Finally, the consensus definition proposed by ECMM/ISHAM, which introduced the use of a lateral flow device or assay (LFD/LFA) as a mycological criterion and serial PCR testing of serum with the aim of increasing specificity, identified 315 CAPA cases (236 probable and 79 possible), the highest number of all of the classifications [106]. However, this was probably because it also includes other kinds of respiratory samples such as BAS or TA that inevitably decrease specificity despite their sensitivity, thus raising doubts about its ability to distinguish real cases of IPA from Aspergillus colonisation. For example, using the ECMM/ISHAM criteria, some of the CAPA cases reported by Alanio, Fekkar, and Rabagliati would be classified as probable CAPA on the basis of only one or two mycological criteria, whereas the authors state that these patients survived without any anti-fungal treatment, thus questioning the validity of the CAPA diagnosis [11,45,61], something that is also confirmed by applying the Verweij and Schauwvlieghe algorithms to the same patients.
The real incidence of CAPA is therefore difficult to estimate, regardless of which clinical algorithm is used. A systematic review by Chong et al., which only considered observational studies using well-defined diagnostic criteria, found an overall incidence of 13.5% (range 2.5-35%) among COVID-19 patients [200], whereas Mitaka et al. found an incidence of 10.2% in ICU patients [201]. CAPA rates probably depend on patient selection and the study setting (mainly ICUs in the studies analysed here), and therefore it is difficult to make meaningful comparisons. For example, studies of ICU patients may suffer from "sample bias" (easier access to lower respiratory tract specimens, enhanced testing in screening programmes for fungal diseases), whereas, in other settings, the limited availability of diagnostic marker testing (PCR testing and serum BDG) and limited awareness of the disease may lead to a significant under-estimate.
Only 62% of the CAPA cases had a positive respiratory culture, and the specimen was BAL in 45% of cases, which means that a large proportion of the diagnoses were based on cultures of less informative respiratory specimens. Only a few studies provided data concerning direct microscopic examinations of respiratory samples, and a positive result (septate branching hyphae suggesting Aspergillus) was reported in only 24 cases (7.4% of the positive cultures: 16 BAL, 5 TA, 1 BAS, and 2 unspecified respiratory specimens) [24,33,35,43,47,52,54,[65][66][67]71,76,[86][87][88][89]92,96,102]. According to Blot et al., positive direct examination of BAL is a crucial criterion for a diagnosis of putative IPA in ICU patients without host risk factors [103]. Given the paucity of direct microscopy data, the diagnostic yield of Blot's algorithm is low, although some authors considered COVID-19 a host risk factor and thus expanded its diagnostic capacity [22].
The diagnostic performance of the Aspergillus-specific PCR testing of respiratory samples was better than that of standard cultures (78.3% considering case reports, case series, and observational studies together). Molecular biology techniques may provide added value when diagnosing IPA, even though most of the data come from adults with haematological malignancies [107], and usefulness of PCR testing in COVID-19 patients has not yet been demonstrated.
The diagnostic performance of serum GM (which ranged from 18% in the observational studies to 53.1% in the case reports; 22.4% overall) was much lower than that of BAL GM (60%), thus confirming the poor sensitivity of the assay in non-neutropenic patients [202]. Moreover, a study by Farmakiotis et al. has shown that BAL GM has a high false positivity rate in patients without haematological malignancies [203]. In 10 studies, GM was tested in respiratory samples other than BAL (BAS, TA, or sputum) and had a positivity ratio of 61.5%; however, in these specimens, GM and PCR testing have not been fully validated, sample collection is not targeted, and it is difficult to establish whether a positive culture corresponds to colonisation or infection and tissue invasion [25,41,49,55]. Serum BDG is apparently diagnostically better than serum GM and has a sensitivity of 36.7% (84/229), but this pan-fungal assay is not specific for aspergillosis.
Making a radiological diagnosis of IPA in COVID-19 patients is challenging for a number of reasons: (1) there numerous technical difficulties in obtaining CT images of patients admitted to an ICU, (2) the radiological features of COVID-19 can mask IPA, and (3) the low frequency of imaging suggesting IPA in mechanically ventilated patients [204]. As only 18.8% of the patients in observational studies and 33.3% of those described in case reports showed imaging features suggesting IPA, it seems that radiological criteria are of little help in diagnosing CAPA.
Finally, it must be remembered that only a few studies attempted to confirm the clinical diagnosis of CAPA by means of HPEs. Only 21 of our reported CAPA cases were histopathologically confirmed, mainly because invasive sampling is frequently unfeasible in COVID-19 patients, especially those who are if critically ill and admitted to an ICU. A recent systematic review of autopsy studies found that the prevalence of autopsy-proven invasive mould disease was only 2% (11/677 deaths, of which 8 were diagnosed as having CAPA [205]), a much lower value than the CAPA rates reported in some observational studies of ICU patients [11,24,30]. As HPE is the reference standard for diagnosing IFIs, the CAPA incidence rates in observational studies are probably over-estimated and the specificity of clinical algorithms is low. On the other hand, autopsies carried out at a single centre in Germany showed four out of eight patients in whom an IFIs was not clinically suspected actually had autoptic signs of an IFI that was considered the cause of death: two cases of CAPA and two of mucormycosis [206].
In brief, the real burden of CAPA has not yet been established as the currently available clinical tools are not sensitive or specific enough to detect it.
The average incidence of candidemia in the observational studies was 3.8%, and the time between ICU admission and a diagnosis was 13.5 days. The latter is similar to the pooled mean duration of ICU admission before the onset of candidemia found by a recent meta-analysis (12.9 days) [207]. Nearly 30% of the case report/case series came from the Middle East, whereas the highest number of observational studies of candidemia came from Italy (33%) and the USA (16.7%). A very high average crude mortality rate was observed in the observational studies (74.8%), but it is worth noting that the ICU outcomes candidemia patients were worse than those observed CAPA patients even before the COVID-19 pandemic [208]. As recently pointed out by McCarty, it is difficult to consider COVID-19 patients hospitalised in an ICU and diagnosed with candidemia as a "truly unique patient population" because most of the patients had classic risk factors (i.e., an in-dwelling vascular catheter, antibiotic treatment, and mechanical ventilation) [209]. However, as 43% received steroids, 19.1% received tocilizumab, and 25.3% were affected by diabetes, it is possible that drug-induced or a metabolic-associated condition was responsible for an increased risk of infection.
The exact global burden of mucormycosis is not known because there is a lack of population-based studies, but its incidence is increasing, with India accounting for most of the cases worldwide [210,211]. It is worth noting that an unusually high number of mucormycosis cases in COVID-19 patients was reported during the second surge of the pandemic in India. Our search identified a total of 3185 cases distributed in 33 case reports/series and 12 observational studies, with an average incidence of 8.7%; many of these came from India (12/33 case reports, 36%, and 7/12 observational studies, 58.3%). Rhino-orbital mucormycosis was the most frequently reported (48.6% overall), and the strong association between COVID-19 and mucormycosis in the Indian population is probably due to a convergence of multiple concomitant risk factors: a higher baseline prevalence of the disease, favourable environmental factors (humidity, soil contamination, air pollution), a high and increasing prevalence of uncontrolled diabetes mellitus, and the widespread use of corticosteroids to treat COVID-19 (often beyond the indications of the guidelines) and possibly other immunosuppressive agents (i.e., anti-IL-6 agents) [176,212]. More than 76% of the candidemia patients considered in this review were affected by diabetes mellitus with poor glycaemic control, and almost 73% received systemic corticosteroids, which may worsen glycaemic control and presumably reduce host immune defences. The characteristics of the candidemia cases (disease localisation, etiologic agents, diagnostic methods, treatment, and outcomes) were similar to those with other IFI [211]. The observational studies indicate a low rate of histopathological diagnoses (43.8%) and positive cultures (23.6%), even though these are required for a proven diagnosis [213]. This is probably due to technical difficulties in obtaining diagnostic samples from COVID-19 patients because of safety concerns or clinical instability secondary to respiratory failure. Furthermore, Mucorales are fragile and, if the material taken from biopsies or other specimen is not managed carefully, a culture may become negative [213]. The treatment of mucormycosis is challenging and requires the prompt administration of anti-fungal agents and debulking surgery but, even after treatment, the average mortality rate associated with the disease is 40% [213]. Among the case reports/case series, maxillofacial surgery was reported in 35 cases (59.3%); however, the outcome was not always favourable probably because of highly invasive disease [140,145,[147][148][149][150][151]157,159,161,163]. More than one-third of the patients in the observational studies underwent surgical debridement. In the largest study of COVID-19-associated mucormycosis, a disease stage of <3b (orbital involvement without vision loss) was associated with a better outcome and, regardless of stage, surgery (paranasal sinus debridement and/or orbital exenteration) was associated with stable residual or regressive disease [177]. Together with infection control, the key elements of successful mucormycosis management suggested in the clinical guidelines are an early diagnosis, species identification, a combination of anti-fungal therapy and aggressive surgical resection/debridement, the optimisation of blood glucose levels, and the correct use of glucocorticoids for COVID-19 treatment (administration to hypoxic patients, and not exceeding the dose and treatment duration established by the guidelines) [213][214][215][216].
A few reports of other IFIs complicating COVID-19 (PCP, cryptococcosis, and histoplasmosis) have been published, but their burden is much less than that of CAPA, candidemia, or mucormycosis. However, physicians should be aware of the immunosuppressive potential of SARS-CoV-2 and consider these diagnoses in patients with suggestive signs and symptoms who experience an unexplained clinical worsening in COVID-19, even in the absence of severe baseline immunodepression [1,19,182]. Diagnosing PCP in COVID-19 patients is difficult and requires a high degree of clinical suspicion as they share some clinical features (signs and symptoms, pulmonary ground glass opacities in CT images, high levels of lactate dehydrogenase (LDH)). Positive Pneumocystis PCR tests of oropharyngeal wash samples or sputum probably over-estimate the real burden of the disease, and therefore treatment with cotrimoxazole may not always be indicated [183].

Conclusions
Increasing published evidence should warn physicians about the association between COVID-19 and IFIs. An increasing number of case reports and observational studies have shown that the clinical course of COVID-19 can be complicated by a variety of fungal super-infections with unfavourable outcomes. Although CAPA, CAM, and candidemia have received the most attention, physicians should remember that the invasive mycoses typically observed in highly immunodepressed patients can also be found in COVID-19 patients. One of the outstanding questions remains the true burden of fungal infections in COVID-19 patients, but the answer can only come from prospective studies that make an extra effort to obtain histopathological confirmation in order to reach a definite diagnosis.

Data Availability Statement:
No new data were created or analysed in this study. Data sharing does not apply to this article.

Conflicts of Interest:
The authors declare they have no conflict of interest.