Oropharyngeal Candidiasis among Egyptian COVID-19 Patients: Clinical Characteristics, Species Identification, and Antifungal Susceptibility, with Disease Severity and Fungal Coinfection Prediction Models

The study aimed to investigate the causative species, antifungal susceptibility, and factors associated with oropharyngeal candidiasis (OPC) among Egyptian COVID-19 patients. This is an observational, case-controlled, single-center study that included three groups: COVID-19 patients (30), COVID-19 patients with OPC (39), and healthy individuals (31). Patients’ demographic data (age, sex), laboratory tests, comorbidities, treatment, and outcomes were included. Candida species were isolated from COVID-OPC patient’s oropharyngeal swabs by convenient microbiological methods. Isolated strains were tested for antimicrobial susceptibility, biofilm production, aspartyl protease, and phospholipase activities. The most common respiratory symptoms reported were dyspnea (36/39; 92.4%) and cough (33/39; 84.7%). Candida albicans was the most common isolated species, accounting for 74.36% (29/39), followed by Candida tropicalis and Candida glabrata (15.38% and 10.26%, respectively). Amphotericin was effective against all isolates, while fluconazole was effective against 61.5%. A total of 53.8% of the isolates were biofilm producers. The phospholipase activity of C. albicans was detected among 58.6% (17/29) of the isolates. Significant variables from this study were used to create two equations from a regression model that can predict the severity of disease course and liability to fungal infection, with a stativity of 87% and 91%, respectively. According to our findings, COVID-19 patients with moderate to severe infection under prolonged use of broad-spectrum antibiotics and corticosteroids should be considered a high-risk group for developing OPC, and prophylactic measures are recommended to be included in the treatment protocols. In addition, due to the increased rate of fluconazole resistance, other new antifungals should be considered.


Study Design, Participants, Specimens, and Data Collection
This is an observational, case-controlled study that was conducted at the internal medicine hospital, Fayoum University Educational Hospital, from 1 April 2020 to 30 June 2021. The study included adult patients (over 18 years) admitted with moderate to critical COVID-19 and confirmed by SARS-CoV-2 RT-PCR from nasopharyngeal swabs. The classification of cases as mild, moderate, severe, and critical was performed as follows: Mild cases have mild symptoms without shortness of breath, dyspnea, or abnormal chest imaging. Moderate cases show signs and symptoms of lower respiratory infection during clinical assessment or imaging and their oxygen saturation was ≥93% in room air at sea level. Severe cases have a respiratory rate greater than 30 breaths per minute, an oxygen saturation of 93%, an arterial partial pressure of oxygen/fraction of inspired oxygen (P/F) greater than 300 mmHg, or more than 50% progression in chest radiological findings within 24 to 48 h. Critical cases have dyspnea that needs mechanical ventilation, shock, or multiple organ failure [27].
During the study period, 69 COVID-19 patients were included in the study, then divided into 2 groups as follows: those who developed OPC were included in one group (n = 39 patients), and the second group included 30 patients without OPC. A negative control group was included in the study and was composed of 31 healthy individuals with negative COVID PCR and no signs of OPC. Patients who developed oropharyngeal candidiasis identified by the presence of pseudomembranous structures or white plaques on the intraoral mucous layer were included in the COVID-OPC group. Oral plaques were sampled using sterile swabs that were plated on Sabouraud dextrose agar for further identification. COVID control group participants were randomly selected from COVID patients who showed no signs of OPC during hospitalization. The negative control group included healthy individuals with negative COVID PCR and no signs of OPC. The collected information included demographics (age, sex), treatment, and outcomes. Data clean-up and manual entry were performed to assure quality. Clinical and laboratory variables associated with the development of OPC and the severity of infection was determined by the comparison of the studied groups based on OPC development and disease severity.

Isolation and Identification of Candida Isolates 2.2.1. Isolation of Candida Species
Oral swabs were plated on Sabouraud dextrose agar (Oxoid, Basingstoke, UK) and incubated at 37 • C for 24-48 h. Isolated colonies were identified using standard microbiological methods (colony characters, Gram staining, urea hydrolysis, and germ tube test). White, creamy yeast-like colonies showed gram-positive budding yeast cells and pseudohyphae under light microscopic and negative urea hydrolysis test were selected for further identification [28]. Selected colonies were examined for candida species by

Phospholipase Assay
Phospholipase activities were investigated among Candida isolates as previously described [29]. Five microliters of the test isolate (10 8 CFU/mL) were inoculated onto egg yolk agar plates and incubated at 37 • C for 48 h. The existence of a precipitation zone around the colony indicates the presence of the phospholipase enzyme. The phospholipase index (Pz) was determined as the ratio of the colony diameter to the total colony diameter and the precipitation zone. When the Pz value is one, the isolate has no phospholipase activity; when the Pz value is less than one, the isolate exhibits phospholipase activity. The lower the Pz value, the more active the phospholipase [29,30]. C. albicans ATCC 10231 was involved as a positive control.

Aspartyl Protease Assay
Candida isolates' aspartyl protease activities were evaluated as previously reported [31]. The degradation of bovine serum albumin (BSA) was measured. Ten microliters of test isolate containing 10 6 CFU/mL were inoculated onto 1% w/v BSA (Levochem, New York, NY, USA) agar plate. The incubation conditions were maintained at 37 • C for five days. Further protease activity was suppressed by adding 20% w/v trichloroacetic acid after incubation (S D Fine-Chem Limited, Mumbai, India), and the plate was stained with 1.25% w/v amido black (Hi-Media, Mumbai, India). Proteolytic activity was demonstrated by a zone of proteolysis encircling the colony that could not be stained with amido black. The protease index (Prz) was calculated by dividing the colony diameter by the diameter of the unstained zone. A Prz value of one indicates no protease activity, while a Prz value lower than one indicates that the tested isolate expresses aspartic protease. The higher the aspartic protease activity, the lower the Prz value [32]. The assay was repeated three times for each isolate. As a positive control, C. albicans ATCC 10231 was used.

Haemolysin Assay
Sabouraud dextrose agar plates enriched with 7% (v/v) blood and 3% (w/v) glucose were used to test the hemolytic activities of candida isolates [33]. The SDA medium was inoculated with (10 µL) 10 8 cfu/mL of Candida isolates. After that, the culture plates were kept for 48 h at 37 • C in a 5% CO 2 environment. Hemolysin production is indicated by the existence of a hemolysis zone around the colony [32]. For each isolate, the assay was repeated three times. As a control strain, C. albicans ATCC 90028 was involved.

Biofilm Formation
The microtiter plate method was used to screen biofilm formation [34,35]. Sabouraud dextrose broth (SDB) (Oxoid, Basingstoke, UK) was used with a final concentration of 8% w/v glucose (Sigma-Aldrich, St. Louis, MO, USA). The candida inoculum was adjusted to approximately 3 × 10 7 CFU/mL. Biofilm production was performed in flat-bottom microtiter plates (Hyundai Micro Co., Seoul, LTD., Korea). Sabouraud dextrose broth (180 µL) was inoculated with 20 µL cell suspension and incubated at 37 • C for 24 h. The plates were rinsed once with distilled water after incubation, and the microtiter plate was stained for 45 min with crystal violet solution. Each well was thoroughly rinsed with sterile distilled water and then destained with ethanol (95%) (Sigma-Aldrich, St. Louis, MO, USA). The destaining solution was quantified using an ELISA plate reader (Euroclone BIOTEK, Pero, Italy) at 595 nm [34,35]. Positive control for biofilm development was C. albicans ATCC 10231. The isolates were classified as non-biofilm producers, weak, moderate, and strong biofilm producers according to their OD590 nm values. IF OD590 nm ≤ ODc the isolate is considered (None-biofilm producers), weak biofilm producers, when OD < OD590 nm ≤ 2 × ODc, moderate biofilm producers,

Antifungal Susceptibility Test
The antifungal susceptibility testing was done using the disk diffusion method in accordance with the Clinical and Laboratory Standards Institute (CLSI) M27-A3 methodology [36]. In the susceptibility tests, amphotericin B (AMB), nystatin (NYS), and fluconazole (FLU) (Oxoid, Basingstoke, UK) were utilized.

Statistical Analysis of the Data
Data was fed to the computer using the IBM SPSS software package version 24.0 (SPSS Inc., Chicago, IL, USA). Qualitative data were described using number and percentage. Comparison between different groups regarding categorical variables was tested using the Chi-square test. Quantitative data was described using the mean and standard deviation for normally distributed data. For normally distributed data, comparison between two independent population was done using an independent t-test, while more than two populations were analyzed using an F-test (ANOVA). The significance test results are quoted as two-tailed probabilities. The significance of the obtained results was judged at the 5% level.

Variables Associated with Oropharyngeal Candidiasis and Severity of Infection
The comparison between the three studied groups regarding demographic, laboratory, and clinical data is displayed in Table 1. Clinical data showed that the heart rate and oxygen saturation were significantly lower in the group COVID with OPC. Regarding laboratory data, most COVID-19 patients with OPC (22/39: 56.5%) showed absolute lymphocytopenia with a significantly lower absolute lymphocyte count (1015.17 ± 634.52 cell/mL) compared to (4/30; 13.3%) with a mean of (1715.33 ± 522.38 cell/mL) in the COVID-control group and (0/30) with a mean of (2777.16 ± 812.98 cell/mL) in the healthy control group (p = 0.001). COVID-OPC patients also showed significant hyperferritinemia (p = 0.001) with a mean value of (788.73 ± 469.22), that is 2 times higher than COVID patients (300.60 ± 196.05) and 5 times higher than the control group (159.16 ± 57.26). Moreover, C-reactive protein levels showed a significant 7-fold increase in the COVID-OPC group (78.16 ± 69.59) compared to the COVID group (11.68 ± 17.22) and 30 times that of the control group (2.55 ± 1.43). Likewise, LDH levels were 2 times higher in the COVID-OPC group with a mean of (539.89 ± 355.94) compared to the COVID and control groups that showed similar LDH levels (236.90 ± 78.75) and (204.32 ± 39.01), respectively. Also, the COVID-OPC group showed significantly higher levels of ALT (91.09 ± 193.26) and AST (98.41 ± 255.62) compared to other groups. Furthermore, albumin, PT, PC, INR, Urea, Ca, and D-Dimer show a significant abnormality in COVID with the OPC group more than in COVID without the OPC and control group. The distribution of selected laboratory test values for the test groups is shown in Figure 1.
OPC group with a mean of (539.89 ± 355.94) compared to the COVID and control groups that showed similar LDH levels (236.90 ± 78.75) and (204.32 ± 39.01), respectively. Also, the COVID-OPC group showed significantly higher levels of ALT (91.09 ± 193.26) and AST (98.41 ± 255.62) compared to other groups. Furthermore, albumin, PT, PC, INR, Urea, Ca, and D-Dimer show a significant abnormality in COVID with the OPC group more than in COVID without the OPC and control group. The distribution of selected laboratory test values for the test groups is shown in Figure 1. (CRP) C-reactive protein, (LDH) Lactate dehydrogenase, (AST) aspartate aminotransferase, (ALT) alanine transaminase. (**) indicate that p-value is less than 0.01 and a significant difference is present; (***) indicates that p-value is less than 0.001 and a significant difference is present. (****) indicates that p-value is less than 0.0001 and a significant difference is present. (ns) mean None significant statistical difference p-value is greater than 0.05. Treatment protocols included the administration of corticosteroids, anticoagulants, antivirals, and antibiotics. Different classes of antibiotics were used for both studied groups, either as singles or in combinations. The commonly used antibiotics included macrolides (AZT: azithromycin), fluoroquinolones (LV: levofloxacin), cephalosporins (CFX: cefotaxime), and oxazolidinones (linezolid). Medications used for the COVID and COVID-OPC study groups are summarized in Table 2. The comparison between the severity of disease and demographics is shown in Table 3, using clinical and laboratory data. It was found that the age was higher in critical cases with high respiratory rate and low oxygen saturation. Laboratory data showed significantly elevated levels of ferratin in critical patients and severe cases compared to moderate cases (p = 0.001). C-reactive protein and LDH levels also showed significant increases depending on case severity, with higher levels in critical cases compared to severe and moderate cases (p = 0.001).

Prediction Models
Multivariate logistic regression analysis was used with significant variables to establish prediction models for disease severity and OPC incidence display significant prediction variables for disease severity (Table 4). These variables were used to create a model of regression to predict the severity of disease from some laboratory items. The equation (Equation (1)) from this model can facilitate the early prediction of the disease severity with 87.0% prediction accuracy. Similarly, our results display significant prediction variables for OPC incidence in COVID-19 patients ( Table 5). The model can predict the incidence of OPC from common laboratory findings with 91% accuracy (Equation (2)).
Disease severity prediction Equation (1) derived from regression model of significant variables. The accuracy of the equation was 87.0%. If (a) is equal or less than one, it is likely to develop moderate COVID. If (a) is greater than one and less than two, it is likely to develop severe COVID. If (a) is greater than two, it is likely to develop critical COVID. (1) (*) mean that p-value is less than 0.05 and a significant difference is present.  (2)

Discussion
The COVID-19 disease course has a wide spectrum of clinical manifestations that vary from asymptomatic or mild cases to critical illness. Disease severity was attributed to several host factors [13,26,37,38]. Early prediction of disease severity course and liability to development of fungal infection will help in personalizing treatment protocols and consequently improve outcomes. Several studies have recently reported COVID-19-related fungal infections, including invasive aspergillosis and oral candidiasis [19,20,39,40]. To date, there is no clear understanding of the pathogenesis and immune response involved in COVID-19 associated fungal superinfection and candidiasis that can explain the currently rising rates. Viral-induced immune-modulation and treatment protocols with extended spectrum antibiotics and corticosteroids were proposed as potential risk factors for fungal co-infection among COVID-19 patients [19,41,42]. In this study, we investigated the patients' variables associated with oropharyngeal candidiasis among COVID-19 patients in a case-controlled setting. We also characterized the causative agents and their antifungal susceptibility pattern.
Immune dysregulation caused by COVID-19 has been suggested as a potential cause of increased oral fungal infection [39,41,43,44]. The result of this study showed that absolute lymphocyte count is a significant variable associated with fungal infection. The COVID OPC group showed a significantly lower absolute lymphocyte count (1015.17 ± 634.52) compared to the COVID (1715.33 ± 522.38) and control groups (2777.16 ± 812.98). Moreover, critically ill patients showed a lower relative lymphocyte count (11.60 ± 8.15) compared to moderate and severe groups (14.40 ± 9.19 and 25.13 ± 11.48), respectively. Our findings are consistent with previous studies reporting lymphocytopenia as a significant abnormality presented in the early phase of COVID-19 that plays a determinant role in prognosis and is associated with higher mortality [1,13,38].
In the same line, Netea et al, reported that T helper 1 and T helper 17 lymphocytes as well as innate lymphoid cells are the main players in protection against mucosal surfaces Candida infections, whereas monocytes, macrophages, and neutrophils are mainly responsible for the protection against systemic candidiasis. [45]. This agrees with the present results as normal monocyte count with an insignificant difference between the studied groups.
On the other hand, absolute neutrophils count was significantly higher in the COVID-OPC group (6267.00 ± 4177.42) compared to the COVID and control groups. Our findings of absolute lymphocytopenia with normal monocytes and neutrophils count are consistent with previous reports [1,24,25,41]. However, it is difficult to explain the liability to fungal infection with the normal or elevated levels of neutrophils and monocytes. A possible explanation for this might be the attenuated monocytes' response toward Candida albicans in COVID-19 patients compared to healthy controls in ex-vivo immune stimulation assays reported by Moser et al. Moser's observation may explain the flourishing of Oral candidiasis despite the elevated levels of monocytes. [41]. Moreover, the study by Melo et al. indicated that elevated neutrophils might indicate potential clinical deterioration and poor outcome. In our study, the critically ill patient showed higher neutrophils levels than moderate cases, which is consistent with Melo et al. findings [13].
The high mortality rate in critically ill COVID-19 patients has been attributed to a hyperinflammatory response and cytokine storm, which cause immune dysregulation and ARDS [7,25]. Early administration of corticosteroids in critical patients reduced 28-day mortality and improved outcomes by controlling hyper inflammation and cytokine storm [11,12]. In this study, most of the patients who developed OPC were on corticosteroids, which is consistent with previous reports showing that oral or systemic corticosteroid administration is a common cause of OPC and fungal infections [46].
Clinically differentiating COVID-19 infection from bacterial and fungal infection is challenging, as both share common symptoms, including cough, fever, dyspnea, and radiologic consolidation [47]. Driven by uncertainty and anxiety surrounding the epidemic, overprescribing broad-spectrum antibiotics has become widespread in the pandemic era [15,47]. Excessive use of broad-spectrum antibiotics is a major risk factor for fungal superinfection due to depletion of the normal flora that creates favorable conditions for fungi to propagate [42]. In the early stages of the pandemic, management protocols included azithromycin for its potential antiviral, anti-inflammatory, and IL-6 lowering properties [25,48]. Other antibiotics such as respiratory fluoroquinolones (moxifloxacin and levofloxacin), cephalosporins, and linezolid are sometimes also prescribed [49,50]. All patients included in this study were treated with one or more antibiotics; most were on a azithromycin/cephalosporin combination. Administration of broad-spectrum antimicrobial may contribute to OPC (Table 2). In a previous study principle, Butler et al. concluded that the routine use of azithromycin for shortening time to recovery or reducing the risk of hospitalization is unjustified. Experts from the Dutch Working Party on Antibiotic Policy concluded their study by recommending antibiotic restrictions for proven or high-likelihood COVID-19 patients who presented with radiological or inflammatory markers consistent with bacterial infection [51].
In the present investigation, C. albicans (74.36%), C. tropicalis (15.38%), and C. glabrata (10.26%) were isolated from OPC patients' oral swabs. Candida albicans was the most prevalent strain in OPC patients, which is in line with previous reports [2,52]. Colonizing oral mucosa with C. glabrata is common in patients receiving radiation for head and neck cancers [53]. Over the past years, infection has increased by 17% and is particularly common in patients with compromised immunity or after prolonged use of broad-spectrum antibiotics [54]. The prevalence of C. glabrata is consistent with Salehi's study. However, C. tropicalis prevalence is five times higher compared to the previous study [42]. This inconsistency can be explained by COVID associated immune dysregulation and the old age of included patients that allowed weak C. glabrata to establish infection without C. albicans. In addition, ICU hospitalization and being bedridden for a long time will enhance C. glabrata infection despite weak adhesion. However, due to the limited number of patients, our hypothesis is not conclusive and patient-controlled study is needed for further investigation.
In this study, isolated species showed high susceptibility levels to amphotericin and nystatin. However, fluconazole showed reduced activity towards different candida strains, and resistant strains were identified among the candida isolates. This can be attributed to the intensive use of fluconazole for the treatment of OPC. Due to immune dysregulation, including lymphocytopenia, monocyte attenuation and cytokine storm, COVID-19 is a potential risk factor for developing OPC. In addition, treatment protocols may increase the risk through extensive use of antibiotics and corticosteroids. Prediction models that can help to predict the severity of disease course and liability to fungal infections will help tailor the treatment protocols as per patient need to reduce unnecessary side effects. These models will support treatment decisions, including the early administration of corticosteroids and antifungals-based severity prediction and liability to develop fungal infection, respectively.

Conclusions
In conclusion, our study highlighted some risk factors for OPC infections among COVID-19 patients. The study has suggested that COVID-19 infection is a potential risk factor for fungal infections. Our data also revealed that COVID-19 patients with moderate to severe infection under prolonged use of broad-spectrum antibiotics and corticosteroids should be considered at high risk of developing OPC, and prophylactic measures should be included in the treatment protocols. In addition, drugs other than fluconazole should be considered due to resistance.