SIMIFF study: Italian fungal registry of mold infections in hematological and non-hematological patients

Purpose We compared the risk factors, the diagnostic tools and the outcome of filamentous fungal infections (FFIs) in hematological patients (HAEs) and non-hematological patients (non-HAEs). Methods Prospective surveillance (2009–2011) of proven and probable FFIs was implemented in 23 Italian hospitals. Results Out of 232 FFIs, 113 occurred in HAEs and 119 in non-HAEs. The most frequent infection was invasive aspergillosis (76.1 % for HAEs, 56.3 % for non-HAEs), and the localization was principally pulmonary (83.2 % for HAEs, 74.8 % for non-HAEs). Neutropenia was a risk factor for 89.4 % HAEs; the main underlying condition was corticosteroid treatment (52.9 %) for non-HAEs. The distribution of proven and probable FFIs was different in the two groups: proven FFIs occurred more frequently in non-HAEs, whereas probable FFIs were correlated with the HAEs. The sensitivity of the galactomannan assay was higher for HAEs than for non-HAEs (95.3 vs. 48.1 %). The overall mortality rate was 44.2 % among the HAEs and 35.3 % among the non-HAEs. The etiology influenced the patient outcomes: mucormycosis was associated with a high mortality rate (57.1 % for HAEs, 77.8 % for non-HAEs). Conclusions The epidemiological and clinical data for FFIs were not identical in the HAEs and non-HAEs. The differences should be considered to improve the management of FFIs according to the patients’ setting.


Introduction
In recent years, the epidemiology of fungal infections has changed. Recent findings indicate an increasing number of filamentous fungal infections (FFIs), most likely due to the widespread usage of fluconazole prophylaxis for Candida albicans control [1,2]. FFIs occur mainly among patients with hematological malignancies, especially during prolonged neutropenia, and in recipients of hematopoietic stem cell transplants (HSCTs) [3]. However, in recent years, the spectrum of high-risk patients has expanded. FFIs have also been recognized as an emerging opportunistic infection in patients with chronic obstructive pulmonary disease (COPD) [4] or connective tissue diseases requiring corticosteroid therapy, in solid cancer patients [5], HIV patients [6], ICU patients [7,8] and patients treated with new immunosuppressive agents [9]. Recently, an increasing number of cases associated with chronic lymphoproliferative disorders has been reported [10,11]. In these non-conventional hosts, FFIs are perceived as less of a concern; thus, the diagnosis is frequently made at a more advanced stage or at autopsy [12][13][14].
Regarding these epidemiological challenges, several studies report a wide distribution of fungal pathogens that reflects marked differences among the patient characteristics and prevention/treatment protocols, as well as environmental characteristics [15]. Indeed, although most FFIs are attributed to the Aspergillus genus, other less common molds, such as Mucorales, Fusarium and Scedosporium spp, are increasingly reported [16][17][18].
The aim of this study was to assess the epidemiology, diagnosis and outcome of FFIs in hematological patients (HAEs) and non-hematological patients (non-HAEs) and to analyze the possible differences between these two groups.

Study design
The Italian Society of Medical Mycology (Federazione Italiana di Micopatologia Umana e Animale, FIMUA) conducted a nationwide FFI surveillance (Sorveglianza Italiana Multicentrica delle Infezioni da Funghi Filamentosi, SIMIFF), enrolling 23 hospitals from January 2009 through December 2011. All new FFI episodes, diagnosed by trained clinicians from mycology and radiology listings and/or histopatology reports, were prospectively collected by the national Coordinating Center (Laboratory of Mycology, Department of Biomedical Sciences and Human Oncology, University of Bari). The diagnostic and therapeutic management followed local practices. Dedicated medical personnel from each unit or hospital were trained in quality-controlled data collection. For each selected case, the hospitals had to complete a report form, including demographic characteristics, underlying disease, predisposing factors, prior antifungal treatment, microbiological, histological and imaging investigations, therapeutic approach and outcome at the 90th day after diagnosis. Every form was sent to the Coordinating Center, where an independent scientific advisory board met periodically to review the data collection procedures and review any diagnosis of infection.
The study was approved by the Ethics Committee of the Azienda Ospedaliero-Universitaria Policlinico of Bari, Italy, and by the institutional review board of each participating center, as appropriate. Registered data were managed in accordance with the Italian data protection laws (privacy law).

Definitions
Only proven/probable cases were enrolled in this study. The classification of each case was defined by the European Organization for Research and Treatment of Cancer/ Mycoses Study Group (EORTC/MSG) [19], with the following modifications: 1. patient enrollment regardless of underlying disease; 2. if diagnosis was made by histology with attainment of large and non-septate hyphae, the case was classified as proven mucormycosis, whereas observation of branched septate hyphae was considered generically diagnostic for hyalohyphomycosis.
In addition, pulmonary aspergilloma (PA) was classified as proven when specific antibodies were associated with a positive culture of Aspergillus species in patients with a solid rounded mass within a pulmonary cavity visualized on a computed tomography (CT) scan. In the absence of positive culture, the patients were classified as having probable PA [20].
The infection was defined as multiple when the patient had evidence of infection in [1 anatomic site and as disseminated when the infection involved [1 non-contiguous site reflecting hematogenous spread. Mixed infection was defined as the infection being caused by different mold genera.
Breakthrough infection was defined as the occurrence of probable/proven fungal infection while on prophylaxis with antifungals generally effective against mold for at least 7 days. Neutropenia was defined as an absolute neutrophil count of \1000 PMN/mm 3 (moderate count, 500-1000 PMN/mm 3 ; severe, 100-500 PMN/mm 3 ; profound, B100 PMN/mm 3 ).
Mortality was considered attributable (AM) to FFIs if the patients died with microbiological, histological, or clinical evidence of an active FFI and if other potential causes of death could be excluded by the responsible physician [21]. The crude mortality was defined as the ratio of deaths to the total number of enrolled patients.

Statistical analysis
Univariate analysis was performed using the Chi square test (Fischer's exact test) and the Mann-Whitney U (Student's t test) when appropriate. Multivariate analysis, using logistic regression analysis, identified variables predicting mortality. The adjusted odds ratios (ORs) and 95 % confidence intervals (CIs) were calculated. Survival curves were prepared using the Kaplan-Meier method, and univariate survival distributions were compared using the Log rank test. In addition, Cox proportional hazard univariate models were performed to assess the survival difference between different FFIs.
Statistical analyses were performed with SPSS 10 for Mac OS X (SPSS Inc., Chicago, IL, USA). All of the tests were two-tailed, and statistical significance was defined as p \ 0.05.
Pulmonary involvement was evident in 94 patients (83.2 %; Table 3), and it tended to be more common among the IA patients than among those with other mycoses (84/86 cases vs. 10/27 cases; p \ 0.001). Disseminated disease was observed in 10.6 % of cases (n = 12; seven aspergillosis and five fusariosis).
Regarding the diagnostic methods (Table 4), the diagnosis was made by microbiological culture in 49 patients (43.4 %) and by histology in six patients (5.3 %). Both methods returned positive results in nine other patients (7.9 %). The galactomannan (GM) assay was positive in 82 IA patients (95.3 %), and this assay produced the only positive microbiological data for 48 patients (58.5 %).
After the FFI diagnosis, antifungal treatment was employed in 109 (96.4 %) episodes, and four patients (three aspergillosis and one hyalohyphomycosis) were not treated because of pre-diagnosis death. Eighty-six patients (78.9 %) received antifungal monotherapy, most commonly with voriconazole (n = 60), liposomal amphotericin B (L-AmB) (n = 16) and caspofungin (n = 7). Combination therapy was administered in 23 cases (21.1 %), generally voriconazole plus caspofungin (n = 16). In addition to antifungal therapy, curative surgical interventions (lobectomies, excisional biopsies and debridement) were conducted in nine patients. The crude mortality rate was 44.2 %. Univariate analysis revealed that the following variables had a significant influence on death (Table 5): profound neutropenia, corticosteroid therapy, and cytomegalovirus infection. In the multivariate analysis, the parameters that were independently associated with an increased risk of death were profound neutropenia (OR 4.6, 95 % CI 1.3-16.6) and corticosteroid therapy (OR 13.6, 95 % CI 3.9-47.5). The 90-day survival rate varied considerably based on the type of infection, and mucormycoses were associated with a worse survival (median failure time for death 8.5 days; mortality rate 57.1 %; Fig. 1b). The HR for mucormycosis compared with invasive aspergillosis was 3.2 (95 % CI 1.6-6.3; p = 0.001).

Non-hematological patients
Among the 119 non-HAEs, lung disease, solid cancer, organ transplantation and trauma were the most common underlying disease/risk factors associated with FFI (Table 1). Other predisposing conditions were prolonged corticosteroid treatment (52.9 %) and prolonged ICU stay associated with ventilation and/or parental nutrition (20.1 %). Diabetes was the sole predisposing factor in eight patients (6.7 %), whereas in another 19 cases (15.9 %), diabetes was combined with other underlying conditions, such as COPD and cancer.
Aspergillosis was the predominant clinical entity ( Table 2) The lower respiratory tract was the most commonly involved site (Table 3), with 83 patients (69.7 %) having only lung infection. Extrapulmonary involvement occurred in 31 patients (26.1 %, mostly eye and paranasal sinuses), and disseminated disease occurred in nine cases (7.5 %).
The diagnosis (Table 4) was mostly performed by positive microbiological culture (n = 97; 81.5 %), of which 36 (37. %) were biopsy samples. In terms of imaging, in patients with IA, pulmonary segmental areas of consolidation were the most frequent finding in CT scans and on plain radiographs. Evidence of pulmonary intracavitary mass suggestive of a fungal ball was observed in patients with aspergilloma and Scedosporium infection. Regarding the ocular mycoses (four fusariosis, two paecilomycosis, two scedosporiosis, one alternariosis and one mucormycosis), slit-lamp examination led to infection suspicion each time that it was performed, revealing chemosis, corneal edema and anterior chamber inflammation. The fundus examination indicated vitreous opacity and white punctate infiltrates. For the patients with cerebral infection, the imaging patterns included hypodense foci and scattered lesions of various sizes, whereas signs of diffuse sinusitis were detected in sinusal disease. Before the diagnosis, fluconazole prophylaxis (mean treatment 11.6 days, range 1-40) was employed only in 10 (8.4 %) patients. An empirical therapy was given in 26 patients (21.8 %; azoles in 15, L-AmB in 9 and caspofungin in two patients).
Combination treatment was used in 14 patients (12.3 %), and L-AmB with an azole was the most frequently administered combination.
The crude mortality rate was 35.3 %. Table 5 shows the variables that significantly affected mortality according to univariate and multivariate analyses. Multiple logistic regression revealed that undergoing mechanical ventilation (OR 6.7, 95 % CI 2.3-19.2) and the presence of multiple infection (OR 4.7, 95 % CI 1.1-18.7) were factors that were independently associated with poor outcome.

Comparison between hematological and nonhematological patients
The results of comparative analyses of clinical and biological findings are reported in Table 6. The distribution of probable and proven FFIs was not similar between the two groups: proven infections occurred more frequently in non-HAEs, whereas probable FFIs were correlated with HAEs. The rate of positive GM antigenemia was higher for HAEs. The only significant difference in the thoracic CT findings was the halo sign, which was more common in HAEs.
The rate of underlying conditions in HAEs differed from that in non-HAEs: neutropenia and cytomegalovirus infection were more common in HAEs, whereas corticosteroid therapy, surgery, diabetes mellitus and mechanical ventilation occurred more frequently in non-HAEs. Fever was a prominent feature among the HAEs.
The proportions of lung diseases, fungemia and infection in multiple sites were similar in both groups, but ocular mycoses were exclusively found among the non-HAEs. No significant difference was detected in mortality between the two groups. HAEs had a shorter median survival time (13.0 vs. 18.5 days), but this difference was not significant. Antifungal prophylaxis and empirical therapy were mainly employed in HAEs, and no significant difference in combination therapy was observed.

Discussion
This study reports the first dataset in Italy that captures and displays information on FFIs in HAEs and non-HAEs. Although this registry cannot yield any information on incidence rate, given the lack of an FFI risk-population denominator, it can be considered representative of all FFI cases in Italy. The first finding emerging between HAEs and non-HAEs is the different distribution of proven and probable FFIs (p = 0.010). HAEs are correlated with probable cases (72.6 vs. 27.4 %), which suggests a different diagnostic work-up. FFI diagnosis was mainly based on clinical and antigen investigations. Notably, the majority of IAs were due to unknown Aspergillus spp (55.8 %) because GM results were promptly obtained in these patients. Therefore, the possibility of obtaining an early and non-invasive diagnosis led the hematologists to neglect pursuing a diagnosis of proven FFI based on invasive diagnostic tools, which are difficult to use in these patients. [24]. In contrast, we observed a high proportion of proven diseases (43.7 %) in the non-HAE patient setting, where the clinicians often do not apply the criteria for probable cases [19] and are encouraged to use a more intensive diagnostic approach to overcome the degree of probability because of the wellknown poor performance of the GM test [25,26]. In fact, the GM levels are significantly more reliable in patients with neutropenia [26], and the low rate of neutropenic patients among the non-HAE group could justify the modest percentage of GM positivity in these patients compared with the HAE patients (48.1 vs. 95.3 %, p \ 0.001). Moreover, prophylaxis with drugs effective against mold can interfere with GM performance [26]. In HAEs, the high yield of GM is most likely because at the time of our survey, posaconazole prophylaxis was rarely performed [27], although recent evidence appears to confirm the importance of the GM assay in the work-up of patients on posaconazole prophylaxis [28]. Currently, the risk of poor GM assay performance in patients under posaconazole prophylaxis [29] should be kept in mind.
Prophylaxis was performed more frequently in HAEs, with a preference regarding the type of azole favoring fluconazole, which may explain the prevalence of empirical therapy in the same patient setting, given the known lack of effectiveness of fluconazole against many Candida spp and molds.
An analysis of the type of infection reveals a clear prevalence of aspergillosis in both types of patients but particularly in HAEs (76.1 vs. 56.3 %, p = 0.102). These results are in agreement with data reported in the literature [12,30] indicating aspergillosis primacy, regardless of the patient type. Although lung diseases (i.e., COPD and cancer) constituted the prevalent baseline condition of the non-HAEs, pulmonary involvement did not differ between the HAEs and non-HAEs (83.2 vs. 74.8 %, p = 0.117), confirming the crucial role of lungs in FFI-risk patients [31].
Overall, the mortality rate was 39.6 %, and no distinct difference between HAEs and no-HAEs (42.2 vs. 35.3 %, p = 0.163) was observed, indicating that the FFIs are severe diseases, most likely because the causative agents have developed strategies for the recognition and/or eradication of the immune defenses [32,33]. For non-HAEs, the present mortality rate was lower than those reported in previous studies [8,10,12], suggesting an increase in clinician awareness of the risk of FFI in these patients.
Regarding non-Aspergillus infections, the data reported in the literature reveal a high mortality rate, most likely because these infections are difficult to diagnose and these fungi have a tendency to be disseminated and resistant to most of the available antifungal agents [34]. In our study, the survival rate varied according to the FFI. A trend toward worse survival was observed among patients with mucormycosis in HAEs and in non-HAEs, although survival in HAEs did not significantly differ according to the type of infection. This tendency might be dependent on the different types of underling clinical conditions of the HAEs, irrespective of the subsequent FFI diagnoses.
Finally, some differences concerning host variables and comorbidities were observed in the two groups. Our findings confirm the importance of the depth and length of neutropenia (particularly correlated with conventional chemotherapy) and of cytomegalovirus infection as a factor contributing to the development of FFIs in HAEs [9,35]. In contrast, corticosteroid treatment was the main underlying condition in the non-HAE cohort. It is well recognized that corticosteroids induce complex immune dysregulation that impairs neutrophil and macrophage function. In addition, the anti-inflammatory properties of glucocorticoids usually blunt the signs and symptoms of FFI, and the patients may not present raised temperatures [36]. These factors may explain the absence of fever in almost half of the non-HAEs.
In conclusion, the FFI epidemiological and clinical data were not identical in the two groups of patients. On the other hand, having diagnosed the FFI cases in the non-HAE patients predominantly on the basis of culture, and in the HAE patients on the basis of GM (as microbiological criteria), it may appear obvious that this different diagnostic management may lead to a presumptive diagnostic delay in non-HAE patients. Nevertheless, any effort should be made to overcome the risk of delayed antifungal treatments avoiding a fever-driven diagnostic approach (enforcing CT scan examination and non-culture antigen-based diagnostics) and using an empirical strategy in case of fever.