Scedosporiosis in patients with acute leukemia: a retrospective multicenter report

Abstract

We retrospectively analyzed 542 proven/probable mould infections registered, in the course of 2 studies, in 8,633 patients with acute leukemia, focusing on scedosporiosis. We aimed to define scedosporiosis incidence and mortality rate over a 15-year period. Only 5 cases of scedosporiosis were identified, all of them involving patients with acute myeloid leukemia (AML). We also reviewed all cases of Scedosporium spp. infections in acute leukemia reported to date in the international literature. The 52 cases analyzed confirmed that acute myeloid leukemia is the category with the highest risk of scedosporiosis. Clinical features of scedosporiosis were extremely variable and closely related to patient immune status. Infection disseminated to multiple sites in a very high percentage of patients and outcome was confirmed to be very poor. In our surveys all patients died, in spite of Amphotericin B compounds or voriconazole administration. Our review of literature found scedosporiosis attributable mortality rate (AMR) to be 77%.In conclusion, scedosporiosis, although extremely rare, represents a big problem for clinicians because of its aggressive clinical presentation and the lack of an effective therapy. New drugs with in vitro activity against Scedosporium spp (voriconazole, posaconazole) should be considered. However, their clinical activity should be more widely demonstrated.

Introduction

Scedosporium spp is a saprophyte mould with worldwide distribution. It is usually isolated from soil, polluted water or sewage. In rare cases it can be responsible for serious fungal infections, particularly in immunocompromised hosts. S. apiospermum (anamorphic form Pseudallescheria boydii) and S. prolificans are the two major pathogens responsible for human infections.1 Despite the growing interest in all fungal infections, little is known about its current epidemiology. Therefore, a retrospective study was conducted to increase our understanding of the epidemiological, clinical and therapeutic aspects of this infection. We will report hereinafter all cases of scedosporiosis we observed over a 15-year period in a large series of patients with acute leukemia, enrolled in two different multicenter surveys in Italy. We also reviewed all cases of scedosporiosis in acute leukemia patients reported so far in the international literature.

We retrospectively analyzed all new cases of mould infections registered in patients with AL between January 1988 and December 1997 and between January 1999 and December 2003 in 2 epidemiological multicenter studies.2–3 The consensus criteria proposed by the EORTC/MSG were used to define invasive fungal infections (IFIs)4 and the analysis was limited to infections classifiable as proven or probable. Every centre which had experienced a Scedosporium spp infection was invited to fill in a form, providing information on the episode.

The following patients’ characteristics were taken into consideration: sex and age, clinical signs, symptoms and site of infection, laboratory data (i.e. neutropenia, neutrophil count <1 × 10/L), microbiologic isolates, treatment received (prophylaxis and therapy), outcome, cause of death, autopsy findings. Death due to scedosporiosis was defined when death occurred in the presence of microbiologic, histologic, or clinical evidence of active fungal infection. In addition, a MED-LINE-based search was performed to identify all reported cases of scedosporiosis in acute leukemia from January 1976 to December 2006. The search terms used were Scedosporium, Pseudallescheria, acute leukemia. Other reports were taken from references of available works. For each reported case, the following data were collected: age, sex, underlying hematologic malignancy, IFI risk factors, geographic location, identified etiologic agent, involved sites, symptoms, diagnostic procedures, antifungal prophylaxis and treatment, and outcome.

We performed univariate (χ test) analysis to identify variables that could predict patients’ death. The parameters we tested as independent variables were: age, sex, type of underlying acute leukemia, dissemination of infection, detected etiologic strain (S. apiospermum vs S. prolificans) and certainty of diagnosis (proven vs. probable IFI).

Retrospective analysis of cases in Italy

During the 15-year period, a total of 8,633 patients with a newly diagnosed acute leukemia (6,303 myeloid and 2,330 lymphoid) were admitted to the participating centers for treatment other than hematopoietic stem cell transplantation. We collected a total of 542 proven/probable mould infections (252 cases in the first study and 290 in the second) for an overall incidence of 6.3% in acute leukemia (5.7% in the first study and 6.9% in the second). Possible cases of IFI were excluded. Data pertaining to the two study populations are summarized in Table 1.

Among all mould infections, only 5 cases (5/542, 0.9%) of proven scedosporiosis were diagnosed in 3 different centres, with an incidence of 0.08% (6/6,603) in acute leukemia. S. apiospermum was responsible for all 5 cases. Patients’ epidemiologic data have been summarized in Table 2. Acute myeloid leukemia (AML) was the underlying hematologic malignancy in all cases. Only one patient developed scedosporiosis in an early stage of the hematologic disease. The remaining 4 were resistant to (3 pts, 60%) or relapsed after prior cytotoxic treatments. All patients had chemotherapy induced severe neutropenia at onset of IFI (mean duration before the onset of IFI: 9 days, range 4–13 days).

Disseminated disease was the most common clinical presentation. Non-specific reported symptoms were fever, cough, and pain. The lung was the most frequently affected site (75% of patients). Other involved sites were sinuses, skin and orbit. Fungemia occurred in 2 patients.

According to the EORTC/MSG criteria, all infections were classified as proven. In two patients histological data were acquired only in the post mortem phase.

Amphotericin B (AmB) was administered in 4 patients, in its deoxycolate (D-AmB) (1 patient) or liposomal (L-AmB) (3 patients) formulation. The remaining patient received voriconazole. This triazole was also administered as a second line therapy in 1 patient, after failure with L- AmB treatment.

Outcome was dramatically poor and all patients died: four of them died due to fungal infection while the fifth patient improved after a salvage treatment with voriconazole but died afterwards for causes other than scedosporiosis. The mean survival from the onset of infection was 17 days. Treatment data are reported in Table 2.

Literature review

The literature search yielded 29 reports of 52 cases of Scedosporium spp infection over the last 30 years in patients affected by acute leukemia.5–33 Cases occurred after stem cell transplant procedures and those published without complete information were excluded. Characteristics of 52 previously published patients are listed in Table 3. Mean age was 47 years (range 3–79), and male/female ratio was 1.9:1. Lineage was not specified in 13 cases, otherwise patients were mostly affected by myeloid leukemia (25/39 cases, 64%) and almost all of them were neutropenic at the onset of infection. Thirty-seven cases (71%) were attributed to S. prolificans.

The geographic distribution of cases was almost exclusively restricted to Spain, with the exception of 8 and 6 cases reported from Australia and USA respectively. Only sporadic cases were found in the other countries. It is worth noticing that almost all cases due to S. prolificans occurred in Spain (27/37, 73%), while S. apiospermum showed a worldwide distribution.

In the majority of cases (43/52, 83%) diagnosis was proven, but in some of them (11/52, 21%) was only obtained in the post mortem phase.

Dissemination of infection (defined as involvement of 3 or more organs) occurred in 40 cases (77%). In particular, scedosporiosis was already disseminated at the onset of symptoms in 14 patients (27%). Other frequently involved sites in the early phase were lung (14/52, 27%), skin (12/52, 23%) and central nervous system (7/52, 13%). Endophthalmitis was the first clinical manifestation of scedosporiosis in 4 out of 52 patients (8%).

The fungal pathogen was isolated from the blood in a large number of patients (29/52, 56%). Only in few cases was it cultured from bronchoalveolar lavage or cerebrospinal fluid. Except for 4 patients, none had received systemic antifungal prophylaxis. In 1 patient, infection occurred during D-AmB administration as prophylaxis. Five patients died before starting a targeted antifungal therapy. D-AmB was the most frequently employed drug as first line treatment, alone (44%) or in combination with azoles (17%). Outcome was extremely poor: 40 patients (77%) died, 7 of them in spite of a salvage antifungal therapy. The presence of disseminated infection predicted a lower survival (AMR 95% vs. 23%, p-value <0.0001).

Table 1.Scedosporiosis observed among moulds infections in 2 cohorts of patients with acute leukemia.

Other variables which influenced outcome of infection were S. prolificans as detected etiologic agent (AMR 89% vs 47%, p-value <0.001) and the certainty of diagnosis (AMR 88% vs 12%, p-value <0.0001).

Some successful cases, after the administration of the new triazoles voriconazole and posaconazole as salvage therapy, have been reported.21, 31

Since the first case of scedosporiosis reported in acute leukemia in 1977,32 hematologists have shown a growing interest in this pathogen, particularly over the last decade. Both our literature review and our case series demonstrated that, among patients affected by hematologic malignancies, those affected by AML are at the highest risk, usually during a period of chemotherapy induced severe neutropenia. However, our retrospective studies in Italy showed that the incidence rate still seems to be very low, even in this category of patients. Despite the limitations of its retrospective approach, this study has collected and made available a large amount of relevant information on all the patients admitted to the participating hematology departments during the study period.

Thanks to the long duration of our survey, we were able to register any change in the IFI incidence rate over the years. As diagnosis of infections has gradually improved, the absolute number of IFI has increased.3

By contrast, the incidence of scedosporiosis remained constantly very low and did not show any temporal clustering. These data do not match the recent report by Lamaris et al.34 In their experience, the incidence rate in cancer patients significantly increased from 0.82 cases per 100,000 patients-inpatient days in the period 1993–1998 to 1.33 in the period 1999–2005. The lack of evidence of nosocomial transmission has led the authors to relate these data to improvements in microbiologic diagnosis or to the selection pressure produced by new drugs particularly active against Aspergillus spp.

Table 2.Epidemiological data of patients with scedosporiosis collected in 2 retrospective multicenter surveys in Italy.

In our literature review, a spatial clustering of cases was evident. Whether the geographic distribution represents a particular ecologic characteristic of the fungus is still to be determined. The proportion of patients with a proven diagnosis was as high in our series (100%) as in the literature data (83%). However, our analytical method does not allow us to firmly conclude that our diagnostic ability has improved, since only a small proportion of cases are published.

Table 3.Table 3A. Literature review: 52 reported cases in acute leukemia (present 5 cases were excluded).

We did not make a proven diagnosis in vivo in every patient and this could explain the low incidence observed. Epidemiologic studies about Scedosporium have been always overshadowed by diagnostic limits and the species is often mistakenly identified as Aspergillus spp.1 Some cases classifiable as breakthrough infections occurring during AmB treatment, have been recently reported35 but only one of the 52 published cases was similar.

Clinical features of scedosporiosis are extremely variable, ranging from superficial and localized forms to disseminated disease. They are closely connected to patient immune status, and the most severe forms are always observed in severely neutropenic patients.36 As previously reported,34 fungemia was common, occurring in 2 out of 5 patients (40%) in our series and in 29 out of 52 pts (56%) in literature. This possibly reflects the Scedosporium ability to sporulate in vivo37 and the high frequency of CVC use. In spite of the coexistence of CVC and scedosporiosis in our series, the low number of cases and the absence of similar data in literature do not allow us to come to any conclusions about this association. The high frequency of positive blood cultures is one of the important features of this mould infection which assists its identification, despite its clinical and radiographic similarity to other IFIs. Other characteristics are the cutaneous involvement with erythematous non-prurigenous skin nodules (often with a necrotic center), the presence of CNS involvement with consequent neurological signs and symptoms, and the high incidence of visual complaints due to endophthalmitis. However, these features do not allow us to make a definite diagnosis since this would require the hystopathologic evidence of hyphal tissue or the evidence of a positive culture from a sterile site.

As demonstrated in a mouse model, S. prolificans isolates are more virulent than S. apiospermum strains, thus showing a positive tropism to the kidney and brain.38 Even if none of our cases were sustained by S. prolificans, infection was disseminated in all patients and each outcome was extremely poor. Similarly, among published cases, there was a high proportion of patients with disseminated disease at the onset of infection (16/52, 31%) and/or during the course of the disease (40/52, 75%).

Despite the improved diagnostic tools and the availability of new antifungal drugs, no progress has been made in treating scedosporiosis in participating centers. Perhaps this reflects the absence of specific guidelines to help clinicians to achieve better results. In fact, at the time of the cases reported in our survey, there was no effective therapy for disseminated scedosporiosis. This was due to the lack of response to AmB, including new formulations.39 Over the last few years, the situation has changed thanks to the availability of new triazoles with a superior activity against S. apiospermum. In particular, in a recent work by Carillo et al.40, voriconazole, posaconazole and ravuconazole were all seen to be active against any strain, with interesting MICs. Our literature review showed that therapeutic options for S. prolificans infection are more limited because of its resistance profile: in vitro susceptibility studies have invariably demonstrated an almost total resistance to AmB and to other available antifungal agents.41 However, voriconazole has shown some promising in vitro activity, alone or in combination with terbinafine.40, 42–43

The routine use of these new antifungal drugs seems to suggest an improved outcome in the future. However, available data is for the moment limited. In fact, present data only concerns a small number of patients and cases are not the specific focus of published reports.

In conclusion, the clinical spectrum of scedosporiosis in acute leukemia mimics that of more common infections caused by filamentous fungi. Although, extremely rare, it still represents a big problem for clinicians, due to the lack of a well-defined optimal therapy and to the severe impairment of clinical conditions of patients at risk. At present, the new triazoles, voriconazole and posaconazole, as well as a combined treatment should be considered as a possible option to modify outcome.41, 44–45 Even though the antifungal drugs registered for empiric therapy (AmB formulations and caspofungin) are not intrinsically active against Scedosporium spp, the very low incidence of scedosporiosis also in AML patients does not justify the use of other antifungal drugs active against these highly pathogenic fungi in the empiric setting.

Footnotes

• Authorship and Disclosures MC and LP co-ordinated the study; MC wrote the paper; MC and LP analyzed the data; LP, CG, MS, GL critically reviewed the paper; all other co-authors collected the clinical data. All authors read and approved the final version. The authors reported no potential conflicts of interest.

• Received May 17, 2007.
• Accepted October 17, 2007.