COVID-19-associated pulmonary aspergillosis in patients with severe SARS-CoV-2 infection

A single-center observational study from Greece

Abstract

INTRODUCTION: 

COVID-19-associated pulmonary aspergillosis (CAPA) is a serious complication affecting patients with severe SARS-CoV-2 infection, and is associated with increased mortality.

OBJECTIVE: 

The objective of this study was to investigate potential risk factors, and to estimate the incidence and mortality in patients diagnosed with CAPA.

METHODS: 

A single-center retrospective observational study was conducted on patients admitted to the intensive care unit (ICU) with severe COVID-19 from October 2020 to May 2022. Patients with deterioration of their clinical status were evaluated with serum galactomannan (GM) for probable CAPA. Baseline demographic patient characteristics, vaccination status, and time period during which each patient was infected with SARS-CoV-2 were obtained, and risk stratification according to underlying comorbidities was performed in an effort to assess various risk factors for CAPA. The incidence of CAPA in the entire cohort was measured, and mortality rates in the CAPA and non-CAPA groups were calculated and compared.

RESULTS: 

Of 488 patients admitted to the ICU, 95 (19.4%) had deterioration of their clinical status, which prompted testing with serum GM. Positive serum testing was observed in 39/95 patients, with an overall CAPA incidence in the entire study cohort reaching 7.9% (39/488). The mortality rate was 75% (42/56) in the non-CAPA group that was tested for serum GM, and 87.2% (34/39) in the CAPA group (P = 0.041). Multivariable Cox regression hazard models were tested for 28- and 90-day survival from ICU admission. An invasive pulmonary aspergillosis (IPA) risk-stratified cox regression model corrected for the SARS-CoV-2 variant of the patient identified the diagnosis of probable CAPA and elevated procalcitonin (PCT) levels measured at least 10 days after ICU admission, as significantly associated with death in the IPA-risk subgroup only, with hazard ratio (HR): 3.687 (95% confidence interval [CI], 1.030–13.199, P = 0.045) for the diagnosis of probable CAPA, and HR: 1.022 (95% CI, 1.003–1.042, P = 0.026) for every 1 ng/mL rise in PCT.

CONCLUSIONS: 

Patients in the IPA-risk subgroup that were diagnosed with CAPA had a lower 90-day survival when compared to patients in the same group without a CAPA diagnosis.

COVID-19-associated pulmonary aspergillosis (CAPA) is an emerging superinfection affecting patients with severe COVID-19.[^1,2] While this complication is still being investigated, several observational studies have suggested potential risk factors for CAPA, such as corticosteroid use, immunosuppression, and preexisting lung disease.[^3-9] The incidence of CAPA varies widely according to study design, population examined, and consensus criteria used to diagnose CAPA.[^10-12]

The diagnosis of CAPA can be challenging, since there is a significant overlap in symptomatology and radiologic appearance between CAPA and severe COVID-19 pneumonia. Furthermore, the characteristic halo sign, a hallmark radiographic feature of pulmonary aspergillosis, is not always evident in patients with CAPA.[^13,14] Instead, detection should rely on recently published guidelines that stress the role of bronchoscopic techniques for the diagnosis of CAPA. In addition, harnessing the use of fungal markers has proven beneficial for the probable and/or possible diagnosis of CAPA. This is important, since timely diagnosis can result in prompt initiation of antifungal therapy, which can potentially reduce mortality.[^15,16]

In this study, we describe the clinical characteristics and outcomes of COVID-19 patients requiring intensive care unit (ICU)-level care who were diagnosed with probable CAPA in a single center in Greece.

Methods

Study design

We conducted a retrospective observational study on polymerase chain reaction (PCR)-positive COVID-19 patients admitted to the ICU between October 2020 and May 2022 in a tertiary-level care university hospital in Thessaloniki, Greece. Patients ≥18 years of age with confirmed COVID-19 infection who were admitted to the ICU for acute respiratory distress syndrome (ARDS) were included in the study. Selective testing with serum galactomannan (GM) was conducted on patients with unexplained deterioration of their clinical status, and those with positive serum GM testing (defined as an index >0.5) were diagnosed with probable CAPA according to the latest European Confederation of Medical Mycology/International Society for Human and Animal Mycology (ECMM/ISHAM) guidelines [Figure 1].[¹⁶] Patients with prior SARS-CoV-2 infection who were admitted to the ICU with ARDS but did not test positive for COVID-19 during the current hospitalization were excluded from the study. Furthermore, patients who had a positive serum GM testing after ICU discharge were excluded from the analysis.

Figure 1: Study design flowchart. ICU: Intensive care unit, GM: Galactomannan, CAPA: COVID-19-associated pulmonary aspergillosis, ARDS: Acute respiratory distress syndrome

Patient-level information was collected from the hospital records, and included demographics, medical history, laboratory tests, and patient follow-up until November 2022. All patients were mechanically ventilated and received dexamethasone 6 mg for 10 days starting from hospital admission according to standard treatment protocol.[¹⁷] Data regarding antifungal therapy were not included in the current study.

To further understand the role of various patient-specific characteristics on disease incidence and course, patients were stratified into three subgroups, according to their underlying medical conditions: invasive pulmonary aspergillosis (IPA)-risk, COVID-19-risk, and low-risk subgroups. IPA risk patients included those with classic risk factors for IPA, such as chronic obstructive pulmonary disease (COPD), asthma, malignancy, and immunosuppression.[^13,18-20] COVID-19 risk patients included those with risk factors for serious COVID-19, such as diabetes mellitus and hypertension (HTN), who did not fulfill the inclusion criteria for the IPA-risk subgroup.[^21,22] The low-risk subgroup was considered the rest of the population tested.

Furthermore, stratification according to the pandemic wave during which patients were infected with SARS-CoV-2 was performed, as an indirect way to estimate the role of the viral strains on the risk of CAPA. Genotypic viral sequencing was not available for the majority of our study population. Nevertheless, we assumed that our patients were infected with the most prevalent variant of each pandemic wave in an effort to provide a rough estimate of the effect of specific viral strains on CAPA risk. Patients diagnosed from the beginning of the study period up to July 15, 2021 (second pandemic wave), were assumed to be infected with the B.1.1.7 strain (Alpha variant); from July 15, 2021, to January 15, 2022 (third pandemic wave), with the B.1.617.2 strain (Delta variant); and from January 15, 2022, to May 31, 2022 (fourth pandemic wave) with the B.1.1.529 strain (Omicron variant).[²³]

Laboratory methods

Nasopharyngeal swabs were processed for the detection of SARS-CoV-2 virus by reverse transcription-PCR from hospital-affiliated certified laboratories. Molecular detection of Aspergillus was obtained through serum GM testing using FungiXpert® Aspergillus GM Enzyme-linked Immunosorbent Assay Detection Kit, Era Biology.

Ethical considerations

This study was reviewed and approved by the ethics committee of the hospital in which the study was conducted (protocol number 61828/20–12/2022).

Statistical analysis

Statistical analysis of the data was performed through IBM SPSS Statistics for Windows, version 26 (IBM Corp., Armonk, NY, USA), and Microsoft Excel. Clinical, procedural, and demographic data are presented as the mean ± standard deviation (SD) or frequencies and percentages as appropriate. Our data were not parametric; thus, categorical differences between the patient groups were evaluated by the Chi-squared test for discrete clinical variables, while differences in paired concentrations were evaluated by the Mann–Whitney U-test. Spearman’s rho correlation analysis was performed to investigate the association between factors with a significance level of 0.05. Cox regression hazard models were then developed by forcing into the multivariable analysis clinically relevant baseline covariates and biomarkers that were univariably significantly associated with the primary outcome. P <0.1 was used for assessing univariable significance. Multivariate Cox regression analysis was performed using the ender method to identify predictors of death. Adjusted hazard ratios (aHRs) are presented along with their respective 95% confidence intervals (CIs). A two-sided P < 0.05 was considered statistically significant. All analyses were conducted through the SPSS v26 (SPSS software, Chicago, IL, USA). Statistical significance was defined as a value of P < 0.05.

Results

Between October 2020 and May 2022, a total of 488 patients were admitted to the ICU for severe COVID-19 ARDS. Out of these, 95/488 (19.4%) patients were tested for serum GM. Positive serum testing was observed in 39/95 patients, with an overall CAPA incidence in the entire study cohort reaching 7.9% (39/488).

Regarding the group of patients who were tested for probable CAPA, the mean age was 63.26 years (±10.235 SD), with 67.4% being male. The majority of patients were diagnosed during the third pandemic wave. Out of the 95 patients, 39 (41.1%) were diagnosed with probable CAPA. Most patients in the CAPA (89.7%) and non-CAPA (76.8%) groups had not received vaccination against SARS-CoV-2. Interestingly, despite vaccine availability, only one patient (2.6%) received a two-dose vaccination regimen in the CAPA group, and only four patients (7.1%) received a three-dose vaccination regimen in the non-CAPA group. The largest subgroup of CAPA patients was the COVID-19-risk subgroup (43.3%), while for the non-CAPA subgroup, most were included in the low-risk subgroup (53.6%). In the CAPA group, 34/39 (87.5%) patients died, as compared to 42/56 (75%) patients in the non-CAPA group, with the difference in mortality rates being statistically significant (P = 0.041). No other difference in demographic or clinical characteristics between the groups was statistically significant.

Continuous variables, including number of days patients stayed in the ICU, 28- and 90-day survival, and death post ICU admission, were assessed in the 95 patients who were tested for probable CAPA, but differences between the CAPA and non-CAPA groups were not statistically significant. Measurements of several biomarkers were collected at the time of ICU admission, and again at various time points since admission (C-reactive protein [CRP] 10 days post ICU admission and procalcitonin [PCT] at least 10 days post ICU admission), to assess disease progression. In addition, random serum cortisol levels that were measured in patients with an unclear cause of worsening of their clinical status were included in the analysis. However, no significant associations were reported between the CAPA and non-CAPA groups with regard to the above variables [Table 1].

Table 1: Demographic, clinical, and laboratory parameters of patients included in the study

In the CAPA group, correlational analyses were performed to find potential variables that were associated with CAPA development and disease progression. Information regarding the strength and direction of the relationship between continuous laboratory variables and CAPA found a significant association between CAPA diagnosis and cortisol levels (correlation coefficient (r) =0.462, P = 0.04). Significant associations were also observed between various biomarkers. In general, an upward trend was recorded for the inflammatory markers studied, indicating the observed worsening clinical course of the study participants [Table 2].

Table 2: Correlations between continuous laboratory variables in the coronavirus disease 2019-associated pulmonary aspergillosis group

Multivariate Cox regression hazard models were tested for 28- and 90-day survival post ICU admission. An IPA risk-stratified Cox regression model corrected for the pandemic wave of the patient identified the diagnosis of probable CAPA and PCT levels obtained at least 10 days post ICU admission, as significantly associated with death in the IPA-risk subgroup only, with hazard ratio (HR): 3.687 (95% CI, 1.030–13.199, P = 0.045) for the diagnosis of probable CAPA, and HR: 1.022 (95% CI, 1.003–1.042, P = 0.026) for every 1 ng/mL rise in PCT [Table 3]. Having CAPA was a significant risk factor for death in the IPA-risk subgroup regardless of the SARS-CoV-2 viral strain. Figure 2 has the survival plot for patients in the IPA-risk subgroup according to CAPA diagnosis. No other parameters were found to be independent predictors of death in any other follow-up period.

Table 3: Cox regression

Figure 2: Ninety-day survival plot for patients in the IPA-risk group. CAPA was found to be a significant risk factor for death in this subgroup. IPA: Invasive pulmonary aspergillosis, CAPA: COVID-19-associated pulmonary aspergillosis, ICU: Intensive care unit

Discussion

In the present study, we described outcomes associated with CAPA among a cohort of 488 patients diagnosed with SARS-CoV-2 infection that required ICU admission. The overall incidence of CAPA in our cohort was 7.9% (39/488). The incidence of CAPA among COVID-19 patients hospitalized in the ICU varies considerably in published studies, from 3.8% to 35%.[¹²] Several reasons might explain this variability. First, the lack of uniform diagnostic criteria, especially during the first waves of the pandemic and before the ECMM/ISHAM guidelines were published, leads to varying incidence rates of CAPA. Furthermore, the limited use of bronchoscopy, and the different samples used to evaluate fungal markers, such as serum, bronchial/tracheal aspirates, and bronchoalveolar lavage fluid, might account for the uneven incidence of CAPA. Differences in the diagnostic performance of each test to distinguish aspergillosis from colonization, and the different combinations of diagnostic modalities used, might account for this variability in published studies.[^7,10,24,25] In addition, lack of awareness of this complication during the early course of the pandemic could be associated with the increasing incidence of CAPA observed later during the pandemic progression. Moreover, since severe SARS-CoV-2 infection is inevitably required for CAPA development, regional differences in terms of the prevalence of specific viral strains, measures undertaken to halt viral dispersion, and patient comorbidities predisposing to severe disease may all explain the variability in CAPA incidence.[¹¹]

In our cohort, genotypic viral sequencing was not performed, and patients were presumed to be infected with the most prevalent viral strain of the pandemic wave during which they were diagnosed with CAPA. Although not accurate, this stratification provided a rough estimate of the effect of specific viral strains on CAPA risk. However, no viral strain was found to be a risk factor for CAPA development. This could be attributed to the limited sample size of our study population, although, up to date, no other study has found a statistically significant association between a specific viral strain and CAPA development.

Next, to assess risk factors that could be significantly associated with CAPA, we divided patients into three subgroups (IPA-, COVID-19-, and low-risk), based on the known classic risk factors for aspergillosis and the risk factors for severe COVID-19. Identifying subgroups of patients with increased risk to develop CAPA could assist physicians to diagnose and treat CAPA earlier and even justify the administration of antifungal prophylaxis in some cases. However, risk group stratification was not significantly different between the CAPA and non-CAPA groups, albeit P value approaching statistical significance (P = 0.056). That is, splitting our cohort into these subgroups showed that once our patients became critically ill with COVID-19, they were all equally susceptible to CAPA even in the absence of classic risk factors for aspergillosis. Our results also indicate that, in the IPA-risk subgroup, having a diagnosis of CAPA, and elevated PCT levels at 10 days, were independently associated with lower 90-day survival.

From our results, sex, age, and COVID-19 vaccination status were not significantly different between the CAPA and non-CAPA groups. However, other observational studies have identified older age, male sex, HTN, and chronic lung disease (mainly COPD), as independent risk factors for CAPA.[^6-9,26-28] A major contributor to CAPA development is corticosteroid use. Results from various published studies have corroborated this association.[^3-5] Since all our patients received corticosteroids, the association between steroid use and CAPA development could not be studied. It needs to be noted though that most patients with COVID-19 treated in the ICU are more likely to receive steroid therapy, especially after the results of RECOVERY trial were published.[¹⁷] Careful interpretation of this observation is crucial, as the significant association between steroid use and CAPA development may be related to severe SARS-CoV-2 infection and not to treatment itself. Finally, our study showed that elevated cortisol levels are correlated with the presence of CAPA, although careful interpretation of this observation is necessary, since the study sample that had available cortisol measurements was small. Elevated cortisol levels have been associated with more severe COVID-19.[²⁹]

CRP, PCT, and ferritin levels were also evaluated as prognostic indicators for CAPA, but differences in results between the CAPA and non-CAPA groups were not statistically significant. However, for both the groups, CRP levels were found to be above the normal limits, which is an expected finding, given the severe inflammatory response triggered by the severe SARS-CoV-2 infection. At day 10 after admission, CRP levels increased in both the groups, with this upward trend likely paralleling the continuous clinical deterioration of our study cohort. PCT levels also followed similar trends, with an overt increase in the CAPA compared to the non-CAPA group at least 10 days after ICU admission, when samples were collected. However, the clinical significance of the latter finding is not clear, due to the limited number of patients that were tested. Ferritin levels were also significantly increased in both the groups in accordance with the rest inflammatory markers.

Diagnosis of probable CAPA relied on the recently published ECMM/ISHAM guidelines that take clinical, radiologic, and mycologic criteria into account. Bronchoscopic biopsy with histopathologic analysis demonstrating invasive fungal growth with associated tissue damage is considered the gold-standard test for CAPA diagnosis.[¹⁶] However, bronchoscopy was not widely used over fears for aerosolization of the virus and procedure-related complications, especially during the early phases of the pandemic. The recent consensus, though, emphasizes the pivotal role of bronchoscopy for the definite diagnosis of CAPA.[¹⁵] In our case, serum fluid GM was used to diagnose CAPA, and bronchoscopy was not performed. According to the guidelines, this approach could only yield a diagnosis of probable CAPA. GM is a fungal cell wall polysaccharide of Aspergillus that is released into the bloodstream and body fluids during active fungal growth in tissues.[³⁰] A positive serologic test for this marker indicates invasive disease, with the reported specificity and sensitivity ranging from 50% to 92.6% and from 94% to 99.6%, respectively.[³¹] The lower sensitivity may be due to testing early in the disease course, before Aspergillus can be detected in the serum, or due to a contained infection in the lungs that does not result in fungal dissemination.[¹⁵] It is important, however, to note that a negative serum GM test does not reliably exclude the diagnosis of CAPA, and, if suspicion is high, further investigational studies are warranted. Detection of GM in serum in case of invasive aspergillosis is usually evident early in the course of the disease.[³¹] In our entire cohort, the median duration from COVID-19 diagnosis to positive serum GM was 22 days. Interestingly, the time to diagnosis was double in the second and third pandemic waves, compared to the fourth one (22.5, 20.5, and 11 days, respectively), indicating the increasing awareness about CAPA as a significant complication in mechanically ventilated COVID-19 patients. Knowledge of when to expect the development of this superinfection is important, since earlier testing for serum GM may lead to false-negative results.

We also assessed the mortality rate in 90 days between the groups. In the CAPA group, 34/39 (87.5%) patients died, as compared to 42/56 (75%) patients in the non-CAPA group, with the difference in mortality rates being statistically significant (P = 0.041). Er et al. also found a statistically significant mortality difference between the groups (67.4% in CAPA vs. 29.4% in the non-CAPA group, P < 0.001), and Hashim et al. also found a higher mortality rate in CAPA versus non-CAPA patients (HR: 1.8 [1.1–2.8], P = 0.001).[^28,32] A systematic review by Chong and Neu included 19 cohort studies on CAPA (6 – prospective and 13 – retrospective), featuring 192 patients. The overall mortality rate was 48.4% (range: 22.2%–100%).[³³] Another systematic review by Mitaka et al. assessed 28 observational studies (9 – prospective and 19 – retrospective) with 297 CAPA patients. The overall mortality rate was 54.9%.[³⁴] The variability on mortality rates may be explained by the sample size of the studies assessing CAPA, the underlying patient comorbidities, the choice of antifungal therapy, and the time period these studies were undertaken, since, early on, CAPA was not a recognized clinical entity, and thus, extensive diagnostic workup was not common practice. Data on antifungal therapy were not included in our study, so we cannot comment on the effect of such therapy on mortality rates. Indeed, most retrospective studies lack information regarding the treatment of CAPA. Mortality rates in the treatment and nontreatment groups are consistently lacking, making inferences about the efficacy of antifungal therapy problematic.

Finally, we observed that, in patients in the IPA-risk subgroup, having a diagnosis of CAPA, and increased PCT levels, obtained at least 10 days after admission to the ICU, were associated with increased mortality. Thus, CAPA screening could be expanded to include COVID-19 patients with elevated PCT levels, although caution is warranted, since PCT levels were measured in a small number of patients.

This was a retrospective study and there was no systematic screening for CAPA, limiting the value of our results. Only patients with disease worsening were tested for probable CAPA with serum GM, which could have underestimated the overall CAPA incidence. Second, genotypic viral sequencing was not performed on our patients. Rather, patients were presumed to be infected with the most prevalent viral strain of the pandemic wave during which they were diagnosed with CAPA in an effort to provide an estimate of the effect of viral-specific strains on CAPA risk. Next, bronchoscopic techniques, which are considered a gold-standard tool for the diagnosis of CAPA, were not implemented in our cohort. This might also have underestimated true CAPA incidence, since serum fluid GM is diagnostically inferior to bronchoscopy. Our sample size was small, which may have prevented us from finding statistically significant associations with regard to risk factors for CAPA development and variables affecting mortality rate. Our study, being single-center, might prevent the generalization of results to other study groups. Finally, data on antifungal therapy were not included in the study, and thus, the effect of treatment on CAPA mortality could not be assessed.

Conclusions

CAPA is a major cause of mortality in patients with severe SARS-CoV-2 infection. Patients presenting with classic risk factors that predispose to IPA have increased mortality rates when infected with CAPA. However, these classic risk factors may be absent in patients presenting with CAPA, necessitating the implementation of established guidelines for timely and accurate diagnosis. A combination of clinical symptomatology, radiographic findings, and mycological evidence of disease through the implementation of bronchoscopic techniques and evaluation of fungal markers can diagnose CAPA in the majority of patients. Suspicion for this disease should be high, since prompt initiation of antifungal therapy can reduce mortality.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.