MSG07: An International Cohort Study Comparing Epidemiology and Outcomes of Patients With Cryptococcus neoformans or Cryptococcus gattii Infections Free

Abstract

Cryptococcus species are saprophytic yeast-like fungi responsible for a broad range of infections in humans. Pulmonary infection results from inhalation of the organism from an environmental source. Cryptococci have a propensity to infect the central nervous system (CNS), and meningoencephalitis is the most commonly recognized manifestation of severe disease [1].

The major species complexes that infect humans, Cryptococcus neoformans and Cryptococcus gattii, vary with regard to geographic distribution, populations affected, disease manifestations, and severity [2, 3]. Cryptococcus neoformans is ubiquitous and abundant in soil contaminated by pigeon droppings. Most often it causes opportunistic infection, typically meningoencephalitis in patients with AIDS, malignancy, or organ transplants and others receiving iatrogenic immunosuppression [1, 4, 5]. Cryptococcus gattii was previously regarded as being limited to primarily tropical and subtropical regions, with a major endemic focus in Australia. It has now been described throughout the world with emergence in Vancouver Island and British Columbia, Canada, and the Pacific Northwest of the United States (US), where it is now endemic, in the late 1990s and early 2000s [6–13]. Ecologically, this species is found in soil and in association with certain species of Eucalypt [3] and other tree species around the world [14–16].

Small studies suggest that C. gattii tends to affect disproportionately patients with apparently normal immune systems compared with C. neoformans [9, 17–19]. Moreover, the frequency of clinical pulmonary disease in C. gattii–infected patients appears higher than in those infected with C. neoformans [19]. Development of intracranial and pulmonary cryptococcomas and need for ventricular shunting may be more frequent among patients with C. gattii infection, but the impact of species vs host characteristics on these disease manifestations and outcomes requires further elucidation [11, 19]. Data informing management and duration of antifungal therapy have either been generated from specific geographic regions, or involved only small patient cohorts, and are now dated [8, 10, 20]. Furthermore, because standard culture methods do not differentiate the 2 species, identification to species level has not been performed in many areas, limiting comparative studies [12, 21].

Recently reported C. gattii infections in the US Pacific Northwest and identification of sporadic human cases in nonendemic areas have stimulated discussion regarding comparisons of these 2 distinct Cryptococcus species. As part of a multi-institutional initiative, we compared risk factors, clinical manifestations, treatment, and outcomes of cryptococcosis among patients with C. gattii or C. neoformans infection in order to inform our understanding of the epidemiology and clinical progression and to guide patient management.

METHODS

Datasets and Study Population

This retrospective cohort study organized by the US Centers for Disease Control and Prevention (CDC) and Mycoses Study Group is based on a master dataset derived from 5 individual site datasets of patients with cryptococcosis. Contributing sites included the University of Alabama at Birmingham (1998–2010) [22]; Duke University Medical Center (1995–2009) [23]; British Columbia Centre for Disease Control (1999–2007) [8–10]; Washington State Department of Health and Oregon State Department of Health (2004–2011) [24]; and Westmead Hospital, The University of Sydney, Australia (1999–2013). Case inclusion required a culture-positive result from cerebrospinal fluid (CSF), blood, other body fluid, tissue, or respiratory sample (sputum, bronchoalveolar lavage fluid) with a compatible clinical and/or radiographic presentation. Species-specific culture methods were used to distinguish C. gattii from C. neoformans cases [25, 26].

Each site used a separate case report form for data collection. Australia included only patients initially diagnosed in the inpatient setting; all other sites included cases diagnosed in inpatient and outpatient settings. Prior to combining data from the 5 individual sites, similarity of case definitions, availability of comparable outcome information, structure of the different databases, and the nomenclature of the respective variables were examined. A final list of variables and definitions was agreed upon by the investigators. These included demographic characteristics, clinical presentation, underlying medical conditions, antifungal treatment, laboratory values including CSF test parameters, imaging studies, and outcomes. To ensure consistency and accuracy in the process of master data set creation, individual-level data for 20 variables from site files were reviewed before final formulation of the master data set. Mortality was defined as all-cause at 3 or 12 months postdiagnosis. Duration of induction therapy of 1 week was defined as 1–9 days; 2 weeks as 10–20 days; 4–6 weeks as 21–49 days; and ≥7 weeks as 50–90 days.

We classified sites of involvement as follows: (1) CNS, which included meningeal and/or parenchymal brain involvement; (2) pulmonary, which included disease limited to the lungs, pleura, and/or pleural fluid; (3) bloodstream, which involved any isolation of Cryptococcus species in blood culture; and (4) other. Disseminated infection was defined as extrapulmonary infection. Patients may have had >1 site of involvement. A patient was classified with no underlying disease if no evidence of human immunodeficiency virus (HIV), transplantation, malignancy, neutropenia, chronic organ dysfunction, diabetes, corticosteroid use, immunosuppressant use, or other immunodeficiency was identified. Malignancy was defined as cancer diagnosed or treated within 6 months prior to diagnosis of cryptococcosis or recurrent/metastatic cancer. Time to diagnosis was defined by the number of days between the date of symptom onset and confirmed date of diagnosis.

Statistical Analysis

Characteristics for the overall cohort and for patients with infection due to C. neoformans or C. gattii were calculated using means and standard deviations, medians and interquartile ranges, or counts and percentages as appropriate. To analyze the relationship between independent variables and Cryptococcus species, univariate analyses were performed using χ ² or Fisher exact test for categorical variables and analysis of variance or Wilcoxon rank-sum testing for continuous variables. Additional analyses of the relationship between lumbar puncture variables or sex and Cryptococcus species were adjusted by HIV infection. Two-sided P values < .05 were considered statistically significant. Logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between Cryptococcus species and outcomes of interest (eg, mortality, complications) with and without adjustment for age, sex, race, time to diagnosis from symptom onset, HIV, chronic lung disease, and site of infection. SAS version 9.3 software (SAS Institute, Cary, North Carolina) was used for all analyses. The study was approved by the institutional review boards (or equivalents) of all sites and the CDC.

RESULTS

In the master data set, 709 (452 C. neoformans; 257 C. gattii) patients had culture-positive cryptococcosis and were included in the final analyses (Table 1). Mean age of patients was 50.2 years; 61.4% were male and 59.0% were white. Cryptococcus gattii patients were older and more frequently female (P < .01 for both). After evaluating the relationship of sex and species and adjusting by HIV, female sex was associated with C. gattii infection (P = .03). Time to diagnosis of cryptococcosis was prolonged in C. gattii–infected patients (mean, 52.2 days for C. gattii vs 36.0 days for C. neoformans; P < .003). Overall, most patients (59.1%) had CNS infections, with lung (42.5%) and blood (24.5%) the next most commonly affected sites (Table 1). Pulmonary infection was more commonly associated with C. gattii infection (60.7% vs 32.1% for C. neoformans; P < .0001). In contrast, CNS, blood, or skin infections were more common in C. neoformans–infected patients (P < .01 for all).

Most patients (78.8%) had underlying disease and in general, those with C. gattii were less immunocompromised; for example, HIV infection (3.5% vs 38.5%; P < .0001) and solid organ transplantation (6.2% vs 25%; P < .001) were less frequent among C. gattii–infected patients, and there was a higher proportion without underlying disease (40.5% vs 10.2%; P < .001). Exceptions were malignancy (19.1% vs 11.7%; P = .007) and chronic lung disease (22.2% vs 8.4%; P = .001), which were more common in C. gattii patients. Among patients with malignancy, only nonhematologic malignancy was more common among C. gattii patients.

Among patients with CNS infection, headache (72.3%), vomiting (37.0%), and altered mental status (36.5%) were frequent clinical symptoms or signs (Table 2). Fever was present in 31% of patients. Symptoms and signs were similar among both patient groups, with the exceptions of altered mental status and nausea, which were significantly more common in C. neoformans–infected patients. In contrast, headache was more commonly associated with C. gattii infection (Table 2).

Computed tomography (CT) and magnetic resonance imaging (MRI) findings are summarized in Table 3. The most common lung abnormalities were mass lesions (56.9%), unilateral (38.3%), and bilateral nodules (22.8%). Multiple lobar infiltrates were more common among C. neoformans–infected patients (13.0% vs 1.1%; P = .002). On baseline brain imaging (either CT or MRI) among 419 persons with CNS infection, 44 (10.1%) had hydrocephalus. Patients with C. gattii CNS disease were significantly more likely to present with brain mass lesions (cryptococcomas).

Lumbar puncture data were available for 419 patients with meningitis (Table 4). Mean or median CSF opening pressure was not significantly different, nor were baseline cryptococcal antigen titers (Table 4). However, mean CSF glucose (13.9 vs 43.5 mg/dL; P < .001) was lower, median white blood cell counts (119 vs 30/mm³; P < .001) were higher, and the percentage of patients with India ink positivity (78.2% vs 71.2%; P = .002) was greater in C. gattii vs C. neoformans infection. After adjusting for HIV, there were no additional significant differences identified (Table 2). In an analysis of CNS infection in patients without HIV (Supplementary Table 1), patients with C. gattii infection had lower mean CSF protein compared with C. neoformans patients (P = .045).

The treatment of cryptococcosis frequently consisted of an amphotericin B (AmB) formulation plus flucytosine for induction therapy (Table 5). For patients with CNS diseases, an AmB formulation plus flucytosine as initial therapy was administered to 76.4% of patients and proportions were similar by cryptococcal species. Lipid formulations of AmB were more commonly used in C. gattii–infected patients (50% vs 39.5%; P = .046). For patients with data available on duration of induction therapy (n = 473), duration was different by species, with a greater proportion of C. gattii–infected patients receiving prolonged (>4 weeks) induction therapy (P < .001).

All-cause mortality was 18.8% at 3 months and 25.5% at 12 months (Table 6). For the 12-month endpoint, mortality was higher among patients with C. neoformans (28.4% vs 20.2%; OR, 1.56 [95% CI, 1.08–2.26]). After adjustment for age, sex, race, time to diagnosis from symptom onset, HIV, chronic lung disease, and site of infection, the strength of association was similar but not significant (OR, 1.46 [95% CI, .86–2.46]). In subgroup analyses of patients with CNS cryptococcosis, 12-month mortality was also greater among C. neoformans–infected patients in crude analysis (26.8% vs 15.9%; OR, 1.94 [95% CI, 1.13–3.34]); but not significant after adjusted analysis (OR 1.99 [95% CI, .91–4.33]). Among patients with pulmonary infection only, mortality was similar between cryptococcal species. The complication of immune reconstitution inflammatory syndrome (IRIS) was significantly less common among C. neoformans patients in crude (3.6% vs 7.1%; OR, 0.49 [95% CI, .24–.97]) and adjusted analyses (OR, 0.28 [95% CI, .10–.82]). There was no significant difference in the proportion of patients with CNS disease receiving permanent ventricular shunt placement (C. neoformans 13.0%; C. gattii 25%; OR, 0.78 [95% CI, .45–1.33]).

Discussion

Cryptococcosis remains an important cause of morbidity and mortality among many patient populations, especially immunocompromised persons. Because routine culture methods and cryptococcal antigen testing cannot distinguish species, it is very likely that cases have been misclassified as C. neoformans in prior epidemiologic studies. This study evaluated a cohort of patients with species-specific culture-positive cryptococcosis in order to compare characteristics of patients with C. gattii or C. neoformans infection. We identified several differences in the epidemiology and outcomes in the 2 groups: Specifically, among C. gattii–infected patients, symptom duration prior to diagnosis was prolonged, patients were less immunocompromised, pulmonary disease was more common, all-cause 12-month mortality was lower, and IRIS was more common.

There has been an increasing interest in C. gattii infections over the past 2 decades due to the emergence of C. gattii in the US Pacific Northwest and western Canada and the identification of sporadic cases across Canada and the US [8, 10–13, 27, 28]. Moreover, previous comparisons of C. gattii and C. neoformans have described that C. gattii infection is more likely to occur in nonimmunocompromised hosts and typically with pulmonary rather than CNS disease [17–19]. It is important to note that C. neoformans and C. gattii produce diseases in both immunocompromised and immunocompetent hosts, although species frequency is different in the 2 general risk groups.

An increased proportion of C. neoformans patients were of male sex. When adjusted by HIV infection, this association was attenuated, but remained significant. Patients differed in frequency of site of disease, with C. gattii patients having a greater proportion of patients with pulmonary disease, and fewer CNS or bloodstream infections. These differences are influenced by underlying diseases in our cohort [3, 17]. HIV-infected patients with C. neoformans infection, compared to other host groups, are more likely to have CNS or bloodstream infection but fewer pulmonary infections [22, 23, 29]. HIV-infected patients in this cohort were much more common among C. neoformans cases.

An important differentiating finding was the prolonged duration of symptoms prior to diagnosis among C. gattii cases. This may reflect a lower diagnostic suspicion among HIV-negative cases, where cancer may have been considered first; closer monitoring of patients with HIV who may be more involved in the healthcare system; or higher proportion of meningitis among HIV cases, leading to more acute presentations. Even though the overall proportion of C. gattii patients who were immunocompromised was less, 60% had an immunocompromising condition. Cryptococcus gattii cases also had fewer CNS and bloodstream infections and a greater proportion of pulmonary infections. Given the higher proportion of pulmonary-only disease among C. gattii patients and the associated nonspecific pulmonary symptoms, cryptococcosis may have been lower in the differential diagnosis and testing for cryptococcosis may have been delayed. Cryptococcus gattii–infected patients were more likely to have underlying chronic lung disease. Perhaps in some cases, cryptococcal infection may have been attributed to chronic lung diseases, delaying diagnosis. A difference in underlying diseases may also affect duration of symptoms, as prior studies have reported a longer duration of symptoms in non-HIV patients with C. neoformans infection [22, 23].

Clinical features among patients with CNS disease were similar. Headache was more common in patients infected with C. gattii, whereas altered mental status and nausea were more common among C. neoformans–infected patients. These differences are not easily explained by the lumbar puncture results, where most CSF parameters were similar, including opening pressure at baseline. However, CSF white blood cell count was significantly higher and CSF glucose correspondingly lower among C. gattii patients. When adjusted for HIV infection, these species-specific differences did not change significantly. Furthermore, there were a greater number of cases of IRIS in C. gattii patients. Elevated white blood cells could reflect a unique feature of the host–pathogen interaction, or simply a greater proportion of nonimmunocompromised hosts with relatively intact inflammatory responses among C. gattii patients. However, among patients without HIV infection, elevated CSF white blood cell counts remained significantly greater among C. gattii patients.

Previous studies have described imaging findings that may help to differentiate C. neoformans or C. gattii infection [10, 18, 30]. Phillips et al studied 152 patients with C. gattii infection, 66% of whom had lung cryptococcomas. Among the 43 patients with CNS disease and available imaging studies, brain cryptococcomas were present in 18.6% and hydrocephalus in 6.9% [10]. Chen et al noted that brain cryptococcomas were more common among C. gattii– than C. neoformans–infected patients [2]. Hydrocephalus is described in up to 50% of C. gattii infections [2, 11, 17–19]. Our findings confirm these observations: First, on brain imaging with MRI or CT, C. gattii–infected patients were more likely to have mass lesions (cryptococcomas) and “normal” imaging was less common. In contrast, on lung imaging, mass lesions were more frequent in patients with C. neoformans.

Current treatment guidelines for CNS cryptococcosis recommend AmB formulations plus 5-flucytosine, with fluconazole recommended for mild to moderate pulmonary disease [1]. In our cohort, patients with C. gattii were more likely to receive lipid preparations of AmB and induction therapy was longer. This may be a result of institutional-specific practice guidelines, where local experience in at least 1 site supports prolonged induction (6–8 weeks) for eradication of disease. In addition, treatment guidelines for nonimmunocompromised patients recommend longer courses of induction therapy and suppression for C. gattii infection, especially those who have large and multiple pulmonary or CNS cryptococcomas [1]. The specific preciseness of recommendations for treatment between C. neoformans and C. gattii cannot be improved by this retrospective, uncontrolled study. However, it supports present recommendations that induction, consolidation, and maintenance therapies are similar for both species and local experience may help guide lengths of this therapy.

Mortality in patients with cryptococcosis ranges from 8% to 50% and is related to underlying disease [2, 22, 31–37]. Typically, immunocompromised HIV-negative patients are older with a variety of underlying diseases and have increased mortality [22, 23, 35, 37]. In this study, C. neoformans was associated with a higher all-cause 12-month mortality. After adjusted analyses, the strength of association remained consistent, but the findings were no longer significant. This is likely related to a decrease in power after multivariable analysis; however, we were not able to collect and analyze data on mortality attributable to cryptococcosis or treatment course, which may further inform the association. The presence of IRIS was significantly less common among C. neoformans patients, even after adjustment for potential confounders.

Our findings should be interpreted in light of several limitations. Our master dataset was derived from several datasets from which patients had been enrolled in overlapping time periods and from different settings (ie, hospitalized, outpatients). Treatment regimens were not standardized, which may have affected overall outcomes. Although we only included variables in the master data set that were defined similarly in the individual data sets, there is a risk of misclassification. Some data (chest imaging, duration of therapy, clinical response, lumbar punctures) were not available from all patients or sites. Finally, for this study we have not considered the newest proposals relating to the further division of Cryptococcus into multiple species, as this discussion has developed since these data were collected [38]. However, a large dataset such as we present herein will be necessary if we are to discern relevant clinical differences, if any, among the newly proposed species classifications for cryptococci [11].

In summary, this description of a large collection of cases highlights differences in epidemiology and clinical outcomes of patients with cryptococcosis due to C. neoformans or C. gattii. In addition, we have identified that cryptococcal species may be associated with mortality. Our work supports the need for species-specific testing in patients with cryptococcosis to better define epidemiologic and clinical patterns, to guide appropriate treatment regimens, and to predict clinical progression and outcome.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Presented in part: 54th Interscience Conference on Antimicrobial Agents and Chemotherapy, 5–9 September 2014, Washington, District of Columbia.

Acknowledgments. The authors thank Gerald McGwin for assistance with the statistical analyses.

Disclaimer. The funder had no role in study design, data collection or analysis, or data interpretation for drafting of the manuscript. The findings and conclusion in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention (CDC). The corresponding author had full access to all the data in the study and final responsibility for the decision to submit for publication.

Financial support. This work was supported by Merck, Inc.

Potential conflicts of interest. J. W. B.: consultation for Pfizer and Eli Lilly; data safety monitoring board for R-Pharm and Viela Bio. S. C.-A. C.: consultation for MSD Australia, Gilead Sciences, and F2G; untied educational grants from MSD Australia and F2G Ltd UK, outside the submitted work. H. O.: grants from CDC. John Perfect: research funding and consultation from Pfizer, Merck, Astellas, Amplyx, Apilli, Martinas, Scynexis, and Minnetronix; advisory board fees for F2G, Merck, Scynexis, Amplyx, Ampili, Minnetronix, and Matinas. P. P.: research grant support from Astella Pharma Canada and Pfizer Canada; training program grant support from Merck. P. G. P.: grant support from Merck, Astellas, Scynexis, Cidara, F2G, IMMY, Mayne, Amplyx, and Appili. All other authors report no potential conflicts of interest.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes