Clin Microbiol Infect
The electronic surveillance system Hema e-Chart allowed us to prospectively collect data and to perform an analysis of invasive fungal infections (IFI) diagnosed in febrile patients as well as the procedures allowing their diagnosis and outcome according to the treatment given. Every patient admitted to 26 Italian Haematology Units with a new diagnosis of haematological malignancy and who was a candidate for chemotherapy was consecutively registered between March 2007 and March 2009. In all, 147 haematological patients with mycoses were identified. Yeasts were found in 23 infections; moulds were diagnosed in 17 proven, 35 probable and 72 possible mycoses. Galactomannan (GM) antigen was the most important test to diagnose probable mould infection; it was positive (cut-off >0.5) in 27 (77%) probable and in nine (53%) proven mould infections. Among patients with probable/proven mould infection who received no prophylaxis or non-mould-active prophylaxis with fluconazole, more patients (n = 26, 78.8%) had GM antigen positivity compared with patients (n = 10, 52.6%) given prophylaxis with mould-active drugs (p <0.05). First-line antifungal therapy was effective in 11/23 (48%) yeast infections and in 37/52 (71.2%) proven/probable mould infections. Twenty patients (14%) died within 12 weeks. The fungal attributable mortality was 30.4% and 17.3% in yeast and proven/probable mould infections, respectively. Among risk factors only age was independently associated (p 0.013) with mortality; sex, underlying haematological malignancy, previous prophylaxis and presence of neutropenia at diagnosis were not significant. A diagnosis of mould infection seemed to have a trend for a better outcome than the diagnosis of yeast infection (p 0.064).
Invasive mycoses are a major cause of morbidity and mortality in haematological malignancies. Even if survival in mould infections is improving [1–4], they continue to represent a serious complication not only during allogeneic haematopoietic stem cell transplant (HSCT) procedures but also during the management of haematological cancers, in particular acute leukaemia. Yeast infections appear to be decreasing , probably because of the wide utilization of azole prophylaxis, but they still remain a significant problem in HSCT recipients [5,6]. These data mainly derive from retrospective epidemiological studies, which refer to patient populations treated more than 5 years ago. The Hema e-Chart Registry [7–10] is an electronic surveillance system that allows the prospective collection and analysis of data regarding febrile events at first diagnosis in a large number of haematological centres. It allows collection of up-to-date information about risk factors and evaluation of the management of fungal infections in current clinical practice. In particular, close monitoring allowed verification of the correct diagnostic procedures of fungal infection and completion of a more accurate follow-up of these patients. Herein we report an analysis of the type of IFI diagnosed in febrile patients as well as the procedures allowing their diagnosis and their outcome according to the treatment given.
Patients and Methods
In the Hema e-Chart Registry every patient admitted to 26 Italian Haematology Units with a new diagnosis of haematological malignancy and who was a candidate for chemotherapy was consecutively registered between March 2007 and March 2009. Data were introduced prospectively into electronic case report forms as previously reported [7–10].
The Hema e-Chart Registry was approved by the Ethics Committee of each participating centre. For each febrile episode baseline characteristics were collected and data were updated every 3 days. The diagnostic work-up was similar among all participating centres and included: cultures from blood and from other sites such as nasal, pharyngeal and rectal swabs, serological markers for invasive fungal disease, and chest computed tomography scan within 4–7 days of fever. Additional tests, such as bronchoalveolar lavage, sinus or brain computed tomography scan, abdominal ultrasound, skin biopsy or fundoscopy, were performed according to clinical signs and symptoms. At the end of the fever episode the investigator specified a diagnosis selecting from among these possibilities: fever of unknown origin; bacterial, fungal or viral fever; fever due to non-infectious causes. All IFI were then centrally reviewed by the scientific Advisory Board and classified according to European Organization for Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) criteria . Failure to respond was defined as progressive or unchanged clinical or radiological parameters compared with the baseline.
Overall mortality included all deaths that occurred within 12 weeks after the start of antifungal therapy. Attributable mortality was considered as death in a patient with documented radiological, microbiological or histological findings suggestive of active fungal infection with no response to treatment and if other potential causes of death could be excluded by the physician responsible for the patient.
Data were analysed by the usual descriptive techniques. Categorical variables were studied by Pearson’s chi-squared test, or by Fisher’s exact test, if needed. Survivorships were analysed using the Kaplan–Meier product limit method followed by a log-rank test in the case of categorical variables, or by a Cox regression in the case of continuous or ordinal variables. Statistical significance was assumed for p <0.05.
In the Hema e-Chart Registry 147 haematological patients with mycoses (72 possible, 35 probable and 40 proven) were identified. There were 91 men, median age 60 years (range 18–84 years). The majority of patients had acute myeloid leukaemia (81.6%) and >90% (133/147) were severely neutropenic (absolute neutrophil count <0.5 × 109/L) at the time of symptom onset. The epidemiological and clinical characteristics of patients are represented in Table 1.
|Age, years (range)||60 (18–84)|
|Acute myeloid leukaemia||120 (82%)|
|Acute lymphoid leukaemia||10|
|Chronic lymphoid leukaemia||2|
|Profound neutropenia (<0.5 × 109/L)||133 (90.4%)|
|Response to first-line antifungal treatment|
|Yeasts (no. of patients):||23|
|Amphotericin B and its lipidic formulations||4/8|
|Moulds (no. of patients, proven/probable):||52|
|Amphotericin B and its lipidic formulations||15/20|
|Fluconazole (empirical treatment)||0/2|
|Deaths due to mycosis in patients|
|With possible moulds||4/72|
|With proven/probable moulds||9/52|
Regarding causative pathogens, yeasts were identified in 23 infections and moulds in 52 probable/proven infections, with a yeast/mould ratio of 1/2.2. In addition, in 72 patients the criteria for a diagnosis of possible IFI of the lung were met. Among yeasts, Candida was the main aetiological agent detected by culture (six C. albicans, five C. krusei, four C. parapsilosis and three Candida spp.) with a predominance of Candida non-albicans; Candida spp. was also present in urine cultures of two severely neutropenic patients with colonization in other sites, and in a patient with hepatosplenic localization. Finally, two yeast infections were caused by Blastoschizomyces capitatus in blood culture and Histoplasma capsulatum diagnosed by gastroenteric tract biopsy.
Among moulds, we found 17 proven, 35 probable and 72 possible mycoses, according to EORTC-MSG criteria. Chest computed tomography scan was suggestive for mould infection in the 72 ‘possible’ patients and in 48/52 probable or proven patients. Four patients had rhinosinusal localization, 10 had disseminated disease.
Proven mould diagnoses were obtained by biopsy in 14 patients (two Aspergillus flavus, two Aspergillus fumigatus, one Aspergillus spp., one Rizhopus spp, eight moulds species), by culture of sterile fluid (A. fumigatus) in one and by blood culture (Fusarium spp. and Acremonium spp.) in two cases. In one case nasal swab was positive for A. fumigatus. Criteria for probable mould infection were present in 35 patients. The GM antigen in the serum was the most important test to diagnose probable mould infection; it was positive (cut-off >0.5) in 27 (77%) probable and in nine (53%) proven mould infections. The β-glucan antigen was positive in one patient. In seven cases moulds grew in bronchoalveolar lavage (two A. fumigatus, one A. flavus, one A. niger, three Aspergillus spp.), and hyphae were found in one cytological examination from bronchoalveolar lavage.
In our study population 83 patients (56%) were on systemic antifungal prophylaxis at the diagnosis of fungal infection. Seven of these had breakthrough yeast infections by C. krusei , C. albicans , C. parapsilosis , Candida spp. , or Histoplasma capsulatum  during prophylaxis with fluconazole (five patients) and itraconazole (two patient). Most of the yeast infections (16/23, 70%) occurred in patients who had not received any systemic prophylaxis. Proven/probable mould infections were diagnosed in 29.7% (19/64) of patients receiving active anti-mould prophylaxis (itraconazole 18 cases and caspofungin one case) and in 39.7% (33/83) of patients not receiving active anti-mould prophylaxis (Table 2).
|Prophylaxis||Possible mould infections||Probable/proven mould infections||Probable/proven mould infections with galactomannan +|
|Mould inactive||36||33||26 (78.8%)|
|Mould active||36||19||10 (52.6%)|
The correlation of the type of prophylaxis with the positivity of GM antigenemia and therefore its possible impact on the diagnosis of mould infection was evaluated. As shown in Table 2, among 33 patients with probable/proven mould infection developed while receiving either no prophylaxis or non-mould-active prophylaxis with fluconazole, 26 were GM positive (78.8%), whereas among 19 patients with probable/proven mould infection diagnosed while receiving mould-active agents as prophylaxis (itraconazole or caspofungin) only ten (52.6%) were GM positive (p <0.05).
The median time from the beginning of fever and treatment initiation was 5 days (range 1–29). The response to first-line antifungal treatment is reported in Table 1. Yeasts failed to respond to first-line therapy in 12/23 cases (52%). Among 52 patients with proven/probable mould infections failure of first-line treatment was observed in 15/52 (28.8%).
All 143 patients were evaluable for mortality within 12 weeks and 20 of them died (14%). Six patients died from yeast infection and one patient from both yeast infection and heart failure. Among patients with mould infections there were 13 deaths, four in 72 possible (5.5%) and nine in 52 probable or proven mycoses (17.3%) (Table 1). None of the patients with infections caused by rare moulds died.
The fungus-attributable mortality was 30.4% and 17.3% for yeasts and proven/probable moulds, respectively. Overall survival of our entire population and of patients with yeast and mould infections are shown in Fig. 1.
The potential prognostic factors for mortality (sex, age, acute leukaemia versus other haematological cancers, antifungal prophylaxis versus not, yeasts versus moulds, neutropenia at diagnosis versus no neutropenia) were evaluated by univariate and multivariate analysis (Table 3) Only age was independently associated (p 0.013) with mortality within 12 weeks from the start of antifungal therapy. A diagnosis of mould infection seemed to have a trend for a better outcome than the diagnosis of a yeast infection (Fig. 1).
|Hazard ratio||SE||z||p >|z|||95% CI|
|Prophylaxis vs none||1.045488||0.5742785||0.08||0.935||0.3562543–3.068157|
|Moulds vs yeasts||0.3695466||0.1989072||−1.85||0.064||0.1286805–1.061269|
The collection of data from the Hema e-Chart Registry allows us to confirm prospectively our previously published retrospective data [1,4] and also to highlight some observations regarding the current clinical practice in the management of fungal infections by haematologists. As the majority of findings regarding mycoses came from studies of selected HSCT patients, they may not be applicable to patients with leukaemia or lymphoma who did not undergo HSCT and had different host characteristics. Our study confirms some well-known epidemiological data. At the onset of mycosis the majority of patients were severely neutropenic, showing that neutropenia continues to be an important risk factor for fungal disease in haematological cancers. Mould infection is a persistent problem in treating haematological patients, above all in acute leukaemia. In fact, in our population >80% of IFI were diagnosed in acute myeloid leukaemia patients, confirming that these patients are correctly defined as at ‘high risk’ for mycosis. In the other haematological malignancies mould infections were more sporadic. The main site of mould disease remains the lung (>90% of the patients), whereas disseminated mould disease was quite a rare event (8%), probably because empirical antifungal therapy and early diagnostic work-up usually scheduled by the Registry participating centres  avoided a disastrous dissemination. The diagnosis of proven mould infection represented a difficult problem also in this prospective survey. Among the 124 patients with suspected aspergillosis, infection was proven in only 13.7% of cases and was considered ‘probable’ according to EORTC criteria in a further 28.2% of cases. In these patients the diagnosis of aspergillosis relied on the detection of GM antigenemia in over 75% of cases (27/35 patients), showing that GM antigenemia is by far the most important test for obtaining a microbiological criterion for diagnosis in routine clinical practice.
Of note, our data further show that the type of antifungal prophylaxis given was correlated with the level of IFI diagnostic evidence. In fact we obtained a GM test positivity in a significantly lower proportion of proven/probable mould infections when mould-active prophylaxis like itraconazole was used. These data indirectly confirm that the positivity of GM antigen could also depend on the administration of systemic antifungal agents effective against Aspergillus such as itraconazole or posaconazole currently used in prophylaxis, as previously published by Marr et al. . Given the key diagnostic importance of GM antigenemia for the identification of probable IFI, as shown also in this study, a reduced sensitivity of GM antigenemia in patients undergoing mould-active prophylaxis may cause an underdiagnosis of cases of invasive aspergillosis, by classifying them as merely possible IFI, potentially leading to insufficient antifungal treatment.
Although cultures were difficult to grow, the rare moulds diagnosed in our population (Fusarium and Acremonium from blood, Rhizopus from bronchoalveolar lavage) were actually identified by cultures, underlining the importance of cultivating every probable fungal material to achieve a diagnosis of species, which can respond differently to various antifungal agents. As in our previous studies [1,13], rare moulds continued to be rare, constituting only 2% of the entire group of mycoses; instead mould spp. were a frequent diagnosis, probably because of the histological difficulty of distinguishing the characteristics of hyphae between Aspergillus and Mucor.
Yeasts represented 15.6% of our mycoses, but they were the largest group of proven infections (57%) (23 yeasts versus 17 moulds) thanks to the possibility of isolating and cultivating them from blood. Yeasts remain a persistent problem in patients with acute leukaemia. Although azole prophylaxis seemed to perform better against yeasts than against moulds, because the majority of yeast infections occurred in patients not receiving prophylaxis, 30% of yeast infections occurred despite prophylaxis, in three cases as the result of resistance of C. glabrata to azoles.
Response to first-line antifungal therapy was better in mould than in yeast infections; in particular the favourable therapeutic response in the mould infections was similar using various antifungal agents (Table 1); this result differs from that reported in a recent study  regarding invasive aspergillosis in haematological malignancies and HSCT recipients, in which the novel anti-mould azoles voriconazole and posaconazole seem to confer a significant advantage, either as primary or salvage therapy, over the lipid amphotericin-containing regimens. However, in that population two characteristics may be taken into consideration: the first being that the patients were retrospectively included between June 1993 and June 2008, a long period during which the new anti-mould azoles and caspofungin were not always available; the second that the population included also allogeneic transplant recipients, in which response to therapy and mortality seem until now worse than in non-transplanted haematological patients .
A third of the patients responded to a second-line therapy in both types of mycoses, irrespective of which antifungal drug was administered. In haematological malignancies the Italian SEIFEM experience has shown a progressive improvement of results in the last decade [1,4] both regarding the response to antifungal therapy and the mortality due to moulds. In yeast infections attributable mortality was comparable to that reported in the literature and in our retrospective study , being similar in both haematological and intensive-care unit patients [16,17], and it has remained constant over the years. On the contrary, also in this prospective study, attributable mortality for aspergillosis seemed to be further decreasing (17.3%) confirming the results of our more recent retrospective study regarding aspergillosis in acute myeloid leukaemia, performed between 2004 and 2007 , in which attributable mortality was reduced compared with the previous study performed between 1999 and 2003 (27% versus 39%) . In our population statistical studies showed that the diagnosis of mould had a more favourable trend for survival than the diagnosis of yeast (Fig. 1).
Decreasing mortality in aspergillosis was recently reported also in a French single institution study ; probably patients treated in more recent years may have had better outcomes because of new developments in diagnosis and early treatment with new antifungal drugs . On the contrary, diagnosis of yeast has remained difficult until now, based mainly on blood cultures, which are positive in only half of patients. This possible delay of diagnosis could translate into a delay also in therapy administration and may explain the reduced efficacy of antifungal treatment against yeasts.
In summary, this prospective study confirmed that patients with acute leukaemia were the most frequent candidates for fungal infections, particularly moulds, more rarely yeast infections also in patients receiving prophylaxis. In current clinical practice diagnosis was obtained mainly by GM antigen and its positivity increased in the absence of systemic mould-active antifungal prophylaxis, which may therefore significantly modify the sensitivity of the diagnostic criteria of probable IFI. In our population no particular risk factors for mortality were identified except age, but a trend was present in favour of higher mortality from yeast infections compared with mould infections. Finally, also this more recent prospective observational study in non-transplanted haematological patients, like our previous retrospective studies, confirmed that mortality due to fungal infections, in particular due to moulds, seems to have fallen over recent years.
The Hema e-Chart project was supported by an unrestricted grant from Merck Sharp & Dohme, Italy. The project was made possible thanks to the support from Roberto Florenzano, Jelena Petrovic and Chiara Scerch on behalf of MSD-Italy. The authors gratefully acknowledge the many dedicated coordinators of the Centres participating in the study.
L. Pagano, S. Cesaro, A. Nosari, G. Rossi P. Viale and F. Aversa constituted the Scientific Advisory Board. M. Caira and M.L. Pioltelli analysed the data. All other authors collected the data.
AN received research support and honoraria from Pfizer, Gilead, Cephalon and Merck. GR received research support and honoraria from Gilead, Roche, Cephalon, Genzyme, Celgene, Shire and Merck. LP and FA received research support and honoraria from Pfizer, Gilead and Merck. MC received research support and honoraria from Shering Ploug, Gilead and Merck. All other authors reported no potential conflict of interest.