The Impact of Corticosteroids on the Outcome of Fungal Disease: a Systematic Review and Meta-analysis

More than 1 billion people are affected by fungal infections (mycosis) each year. Well over 2 million infections are life-threatening, especially in immunocompromised patients [1], for example individuals with solid organ or stem cell transplants. This population has expanded due to medical advancements, increasing the incidence of significant fungal infections [2]. Immunosuppression as a risk factor highlights the crucial function of the immune system in controlling opportunistic fungal infections. In order to combat this, increasing host immune function or targeting interactions between the host immune system and fungus might be used in conjunction with antifungal medications [3].

Corticosteroids are commonly used in immunocompromised patients. They are a risk factor for development of invasive fungal infection (IFI) [4] and chronic pulmonary aspergillosis (CPA) in patients with pulmonary nontuberculous mycobacteria [5, 6]. Several studies reported that the early use of corticosteroids for acute graft-versus-host disease is a key risk factor for IA [7, 8, 9, 10]. Additionally, topical steroid is regarded a major risk factor for the development of fungal keratitis [11]. Recently, many studies reported that corticosteroids were the risk factor for COVID-19-associated pulmonary aspergillosis (CAPA) [12, 13, 14, 15]. Corticosteroids limit vasodilation, boost capillary permeability (humoral response) and leukocyte migration to wounded tissue (cellular response). They exert most of their immunosuppressive and anti-inflammatory effects via the glucocorticoid receptor, which inhibits the activity of critical transcriptional regulators of pro-inflammatory genes, including NF-κB, in leukocytes. In addition, corticosteroids decrease the number of monocytes and macrophages in circulation by inhibiting their myelopoiesis and bone marrow release. Notably, glucocorticoids suppress phagolysosomal fusion in macrophages by stabilising lysosomal membranes during phagocytosis. Due to the inhibitory effect on phagocytic function, there is an immediate danger of infection with high-dose glucocorticoid therapy [16], especially in HSCT recipients and patients with autoimmune diseases such as systemic lupus erythematosus.

Nowadays, corticosteroids may be necessary to control an underlying disease even when an IFI occurs. It is usually a major clinical dilemma as to whether to and to what extent to withdraw their use. Furthermore, corticosteroids are advocated for some fungal infections as adjunctive therapy, notably PCP in AIDS patients. No comprehensive systematic review has summarised the outcome of the corticosteroids in patients with fungal disease except a meta-review of PCP [17], which only focused on randomised controlled trials. Our systematic review assesses fungal disease patients’ outcomes (survival, vision) with corticosteroid treatment.

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed throughout the whole process of conducting and presenting this systematic review. This study did not require ethical approval or informed consent because it was a systematic review of previously published studies. The study was registered in INPLASY (INPLASY202280110) in August 2022.

In the initial search, 2882 studies were identified. After removing duplicates and papers not meeting the inclusion criteria, 59 studies satisfied the inclusion criteria and were included in this meta-analysis (Fig. 1). These studies concerned different fungal diseases including 19 PCP studies, 18 of invasive candidiasis, hepatosplenic candidiasis and candidemia, 11 of invasive aspergillosis and CPA, three of mucormycosis, three of cryptococcal meningoencephalitis, two of fungal keratitis, two of coccidioidomycosis and one of fusariosis. A summary of these studies is provided followed by a meta-analysis.

This systematic review and meta-analysis is the first systematic evaluation of the outcome of serious fungal disease with corticosteroid treatment. Our review includes various fungal infections in various countries, including randomised controlled trials and observational studies. Most of the research in our analysis pool are retrospective cohort studies as corticosteroid treatment has been confirmed as the risk factor for many fungal infections, so planning and executing RCTs on patients with or without corticosteroid to possibly demonstrate a worse outcome would be unethical in most circumstances.

We analysed patients’ outcome (mortality and vision change) linked to the effect of corticosteroid treatment as a discrete analysis. Previously corticosteroids have been assessed as a risk factor for development of a serious fungal disease, and this is so well accepted that it is mentioned in almost every article written about life-threatening fungal infection. Additionally, we analysed each fungal infection independently to avoid any bias or clouding of contrasting outcomes caused by different pathogenesis and the basic therapy regimen of fungal disease.

The meta-analysis of seven randomised controlled trials in HIV-infected patients with PCP found that the risk of death reduced to 62% (p = 0.001) in those treated with corticosteroids. Notably, the trials were in moderate and severe PCP patients, with a more substantial effect on worse disease. Although Montaner et al. [22, 83] had less severe patients in their studies, the effect of corticosteroids on survival could not be evaluated because of the crossover design of these two studies. There was insufficient evidence to conclude that supplementary corticosteroids affected mortality in babies with PCP and HIV (although it is likely). Evidence from RCTs for HIV-infected patients with mild PCP is still lacking. All these RCT studies were done in high-income countries with low Mycobacterium tuberculosis (TB) rates and were conducted 25 or more years ago. Other endemic infections in specific countries and populations may impact the value of corticosteroid treatment since uptake of highly active antiretroviral (HAART) and trimethoprim-sulfamethoxazole therapy is variable [84]. The value of corticosteroids for PCP in AIDS is of uncertain overall value in countries with high rates of tuberculosis.

Aspergillus spp., filamentous fungi, are frequently isolated from soil, organic matter, food, the indoor environment and hospitals [85, 86]. In our meta-analysis, all 11 cohort studies showed that steroid therapy had a worse outcome for both invasive aspergillosis and CPA. Notably, we found a major increase in death in IA of 250% (p < 0.001) in our pooled analysis and 270% in CPA. This is the first systematic review of this literature. Our review includes different underlying conditions linked with IA including chronic obstructive pulmonary disease (COPD) and coronavirus disease 2019 (COVID-19)-associated pulmonary aspergillosis (CAPA). All of them showed the same results in patients treated with steroids. Like our analysis, Raghu et al. presented a large cohort study from India (abstract only) and found that higher cumulative steroid dosage and longer duration of corticosteroid treatment are independent risk factors for both the development of CAPA and an increase in CAPA mortality [87]. Therefore, we recommend careful adherence to the existing steroid guidelines for COVID-19.

However, there are some limitations of our analysis of aspergillosis. First, all included studies were retrospective studies; some records may have bias and detailed laboratory records to confirm diagnoses were unavailable. In addition, our analysis of publication bias by funnel plot showed the possibility of bias (Supplementary data). The reason for this may because some small studies were not published. Moreover, researchers might prefer to report positive results showing higher mortality; negative or inconclusive studies might have been ignored. Although our analysis found that more than 2 mg per day per kg of steroid (prednisolone equivalents) use was a risk factor for death, this does not mean that less than 2 mg/kg/day of steroids is safe, due to lack of evidence.

Candida species are a significant source of hospital-acquired bloodstream infections and can cause serious infections linked with extended hospital stays and high fatality rates [88, 89]. Our meta-analysis showed that steroid treatment was a risk factor for mortality in candidemia and invasive candidiasis patients (213% increase, p < 0.001). However, there was no significant difference between steroid and non-steroid groups in the rare group with hepatosplenic candidiasis. Ours is the first study to integrate different observational studies of candidiasis with steroid treatment and outcome.

As for hepatosplenic candidiasis, Jang et al. [64] demonstrated no statistically significant differences in 90-day mortality between hepatosplenic candidiasis patients receiving corticosteroid medication and those who did not. Furthermore, Legrand et al. [54] showed that adjuvant corticosteroids reduced clinical symptoms and inflammatory responses in ten patients with hepatosplenic candidiasis, while Chaussade et al. reported rapid improvement in this disease’s symptoms in five patients with hepatosplenic candidiasis who were administered corticosteroids [90]. However, these earlier investigations did not identify the proportion of patients with hepatosplenic candidiasis who required corticosteroids due to incapacitating persistent fever, nor did they describe the clinical criteria that distinguished these patients from those who did not require corticosteroids. There is evidence that hepatosplenic candidiasis is a form of IRIS, and the corticosteroids or anti-inflammatory medications may benefit certain patients. Therefore, further studies should continue to explore the role of corticosteroid treatment in hepatosplenic candidiasis patients.

Long-term glucocorticoid use contributes to mucormycosis risk [91]. In our meta-analysis, we identified three studies which reported that corticosteroid treatment was a risk factor for death in patients with mucormycosis (increased risk of death of 419%, p = 0.001). Notably, mucormycosis pre-COVID-19 has a terrible prognosis, ranging from 33.3 to 80% overall mortality [92]. The limitation of our systematic review of mucormycosis is the small size of the three studies (total patients: 164) making it problematic to generalise our results, although the size of the effect is large. Additionally, two studies were related to COVID-19 patients while only one focused on other in-hospital patients, which may lead to a selection bias and not reflect the general effect of steroid treatment for mucormycosis outcome.

Fungal keratitis is challenging to identify and cure. In addition, it is frequently mistaken with other causes of infectious keratitis due to a lack of clinical and microbiological evidence in its early stages, resulting in delayed treatment. Very soon after corticosteroids were introduced in the 1950’s Thygeson warned that they might be significant factor in worsening fungal keratitis [93]. Since then, only one cohort study has addressed this concern and the published data indicate that recurrent overall treatment results were poorer in patients treated with topical steroids [76]. In contrast, topical corticosteroids after corneal grafting appear to be beneficial in preventing rejection. Topical corticosteroids are presently regarded as the optimum therapy for avoiding early immunological rejection and managing inflammation following keratoplasty [94, 95, 96].

Our analysis of topical corticosteroids and fungal keratitis has some limitations. First, only one study addressed the negative effect of steroid treatment in the early phases of fungal keratitis, while the diagnosis is being considered; and the other only addressed corneal graft rejection. The study in South Korea was conducted at a single tertiary hospital. Perhaps these findings cannot be generalised.

Rajasingham estimated a mean worldwide cryptococcal antigenaemia prevalence of 4% (95%CI: 16–74) among HIV-positive individuals with CD4 counts of fewer than 200 cells/L, equivalent to 179 000 (IQR 133 000–219 000) cases of cryptococcal antigenaemia globally in 2020. Annually, 152 000 cases (111 000–185,000) of cryptococcal meningitis were reported, resulting in 112 000 deaths attributable to cryptococcal disease (79 000–134 000). Cryptococcal illness is responsible for 19% (13–24) of AIDS-related deaths worldwide [97]. Our systematic review found three studies related to cryptococcal meningitis with steroid treatment, with contrasting results. A clinical cohort study demonstrated that steroid therapy dramatically decreased blindness in immunocompetent subjects with infection caused by C. gattii complex [73]. However, Beardsley et al. [74] found that steroids did not reduce mortality among patients with HIV-associated cryptococcal meningitis, but also were not harmful overall. Recently Anjum et al. [75] demonstrated that pulse corticosteroid therapy is associated with improved visual field outcome, and the patients they chose were previously healthy patients with the post-infectious inflammatory syndrome following cryptococcal meningitis, similar to Seaton’s study group. As in PCP, there appears to be a dichotomy between HIV and non-HIV infected patients: with a reversal of benefit in non-HIV. In non-HIV patients without immunosuppression, there is the likelihood of benefit. The main limitations of the data related to non-HIV patients are the study design, timing and dose of corticosteroids and relatively small number of patients. A randomised study could explore the dose of steroids in patients with cryptococcal meningitis without immunosuppression, although an adaptive design should be adopted to reduce the number of participants who are exposed to the inferior arm (whichever that is).

Coccidioides immitis and C. posadasii are sibling species of endemic fungi indigenous to the Americas, notably southwestern deserts of the USA and other arid areas in Mexico and further south [98]. Our systematic review focused on mortality, and we found two studies that reported the effect of corticosteroid treatment on the outcome of coccidioidomycosis. Blair’s [79] 16-year study comprised 55 patients, with the majority of infections (43/55, 78%) occurring in the lungs. According to the study, 34 patients used corticosteroids, and 16 individuals died. However, the number of deaths in the steroid group was not specified. They discovered that steroid use was associated with increased mortality along with hematologic malignancy (p = 0.02). On the other hand, Azadeh et al. identified 74 patients with primary pulmonary coccidioidomycosis [80]. They discovered that corticosteroid medication might assist immunocompetent patients with coccidioidomycosis in recovering faster. In addition, amphotericin B deoxycholate (AmBd) is used to treat coccidioidal meningitis but the majority of patients who receive intrathecal AmBd will have headache, nausea and vomiting [99]. Notably, concurrent intrathecal injection of corticosteroids, such as hydrocortisone or methylprednisolone, reduces the toxicity of the drug in the spinal canal, without apparently prejudicing outcome [100]. A previous review [101] concluded that long-term glucocorticoids might have deleterious effects on coccidioidomycosis patients. We found the opposite result in that steroid treatment may help to relieve symptoms faster in primary infection. However, it is difficult to conclude this, as the sample size is too small. As for bacterial meningitis, such as tuberculous meningitis, some researchers have found that adjunctive corticosteroids reduced death when compared with antibacterial treatment alone [102, 103]. Further study should extend the research time and increase the number of patients. Furthermore, corticosteroids’ influence on the clinical course of coccidioidomycosis, including coccidioidal meningitis, warrants more investigation.

Fusarium species cause a variety of human illnesses, including superficial, locally invasive and disseminated infections. Invasive fusariosis is a particular problem in neutropenic patients, with over 90 documented cases [104, 105, 106, 107]. In our meta-analysis, we identified one study which reported that corticosteroid therapy was a risk factor for disseminated fusariosis. The authors also revealed that corticosteroid therapy led to a poor outcome for patients with fusariosis. Regarding the clinicopathologic characteristics of fusariosis in individuals with hematologic disorders these data confirm and expand the findings of earlier large series [108]. Other studies have addressed the prognosis of fusariosis, but steroid therapy was not included as a mortality factor.

Disseminated histoplasmosis (DH) is a persistently progressive granulomatous illness caused by Histoplasma capsulatum, an intracellular dimorphic fungus [109]. Although we did not find any papers about the relationship between the mortality of DH and corticosteroids treatment, a previous case report and literature review found that a high dosage of corticosteroids was the most prominent factor for deadly infection [110]. In addition, a recent study found that corticosteroid use was an independent predictor of DH infection [111].

Our meta-analysis provides a comprehensive overview of the published data linking outcome and corticosteroid treatment in patients with different fungal diseases. However, it has several limitations. First, most studies had a retrospective cohort design, possibly with inadequate data on confounding factors. There may be a selection bias in the studied predictors, as different doctors collected the detail on each case, and missing data plagues such research efforts. Second, there were no natural control groups in some cohort studies as these studies addressed many aspects of the disease not only outcomes. Some biases in data selection and analysis are likely. Third, the dosage and timing of corticosteroid therapy were not included in most papers, except for aspergillosis. It is likely that both dose and duration are important in raising both the risk of fungal disease and affecting outcome. The current published data cannot address this meaningfully. This makes it difficult to advise clinicians on minimising adverse outcomes based on corticosteroid dosing. Fourth, our study did not include the effect of inhaled steroids in lung fungal infection because there has been little study of inhaled steroids and patient survival in complex clinical settings. Lastly, we found that some fungal diseases (allergic fungal rhinosinusitis, talaromyces and dimorphic fungal disease as examples) do not report any outcome of corticosteroid treatment. We did not include allergic bronchopulmonary aspergillosis in our analysis. We have not addressed the likely impact of azole/corticosteroid interactions, which may have acted to effectively boost corticosteroid exposure, in those taking prednisolone and voriconazole and methylprednisolone and itraconazole, as examples. Moreover, no study reported renal transplant loss related to withdrawal or not of corticosteroid therapy in the context of life-threatening fungal infection [112]. Future research could focus on the impact of corticosteroids on the outcome in renal transplant recipients and explore if reduction in steroid dose could have a positive impact on retaining renal allografts, without the patient dying.

Our review demonstrates that with a few notable exceptions, corticosteroid therapy worsens outcome in several fungal diseases. The inter-relationship between corticosteroid dose, duration and outcome was more difficult to tease apart. Where the data is weak and somewhat equivocal, randomised or careful case-controlled studies should examine the impact of corticosteroid dose on the prognosis of various fungal infections.