1 Introduction

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is an effective and sometimes the only curative therapy for patients with high-risk hematologic malignancies. Due to the therapies preventing and treating graft-versus-host-disease (GVHD) and cytopenia, HSCT recipients have a high risk of acquiring invasive fungal infection (IFI) (Jantunen et al., 1997; Bjorklund et al., 2007).

Over the past decades, multiple changes in transplant procedures and antifungal prophylaxis have affected the epidemiology of IFI (Neofytos et al., 2009; Kontoyiannis et al., 2010). Although fluconazole prophylaxis has reduced the incidence of IFI caused by Candida albicans, it resulted in a subsequent increase in resistant candidiasis, such as Candida glabrata and Candida krusei. Aspergillus has become the most frequent organism and other infections caused by molds, such as Zygomycetes and Fusariums, also have increased since the 1990s (Neofytos et al., 2009).

Previously identified risk factors of IFI include donor type, GVHD, high-dose corticosteroids, increased ferritin levels, and so on (Fukuda et al., 2003; Thursky et al., 2004; Ozyilmaz et al., 2010; Omer et al., 2013). Epidemiology and risk factors of IFI differ due to the changes in antifungal prophylaxis and treatment in different regions. Exploring these differences may improve management strategies and study design for allo-HSCT recipients.

Here in our study, we retrospectively reviewed 408 consecutive patients who underwent allo-HSCT between 1998 and 2009 in our center, identified the incidence and risk factors of IFI, and calculated the outcomes.

2 Subjects and methods

2.1 Patients and donors

We collected a total of 408 patients, 251 males and 157 females aged 8–52 years (a median age of 28 years), who underwent allo-HSCT in our center between November 1998 and December 2009. According to the French-American-British (FAB) criteria, 127 (31.13%) cases were acute myeloid leukemia (AML), 120 (29.41%) were acute lymphocytic leukemia (ALL), 127 (31.13%) were chronic myeloid leukemia (CML), and 34 (8.33%) were other hematologic diseases. Among the allogenic HSCT recipients, 195 (47.79%) received bone marrow transplantation (BMT), whereas 213 (52.21%) received peripheral blood stem cells (PBSCs) with a granulocyte colony-stimulating factor (G-CSF) mobilized. Donors were human leukocyte antigen (HLA)-matched (n= 298), HLA-mismatched (n=82), or HLA haploidentical-matched (n=28). Clinical characteristics of the entire cohort are presented in Table 1.

Table 1 Characteristics of 408 allo-HSCT recipients *
Full size table

2.2 Transplant procedure

A total of 349 (85.54%) patients received myeloablative conditioning regimens composed of busulfan (BU) plus cyclophosphamide (CY). The 27 (6.62%) patients who received HLA haploidentical transplants received BU/CY/cytosine arabinoside (Ara-C)/1-(2-chloroethyl)-3-(4-methylcyclohexyl)-1-nitrosourea (methyl-CCNU)/anti-thymoglobin (ATG). Reduced-intensity conditioning regimens which were predominantly fludarabine-based combinations were used in 32 cases. In addition, 66 (16.18%) individuals with HLA-mismatched donors received ATG as part of their conditioning regimen.

2.3 GVHD prophylaxis and therapy

In both unrelated and sibling transplantation cohorts, patients received the same GVHD prophylaxis regimen consisting of cyclosporine, mycophenolate mofetil, and short-term methotrexate (MTX). Acute GVHD was graded from 0 to IV according to established criteria. Patients who survived for >100 d were determined as chronic GVHD, which was graded as limited or extensive. Acute GVHD was treated with high-dose methyl-prednisolone (MP), and ATG was used for patients refractory to MP.

2.4 Supportive care

All patients were cared for in rooms with high-efficiency particulate air filters. Empiric antibacterial agents for fever and cotrimoxazole as pneumocystis prophylaxis were administered. For the prophylaxis of cytomegalovirus (CMV), ganciclovir or foscarnet sodium was used before transplantation and acyclovir combined with immunoglobulin after transplantation.

2.5 Definition, prophylaxis, and therapy of IFI

IFI was defined as proven or probable according to the European Organization for Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) criteria (Ascioglu et al., 2002). Proven fungal infection required histopathologic findings from biopsied tissues or culture of sterile tissue. Probable infection was considered when patients had both clinical criteria and at least one microbiologic criterion (fungus was identified from sputum or bronchoalveolar lavage fluid). Infections were classified according to occurrence timing after transplantation: early IFI (≤100 d after transplantation) and late IFI (Group I: 100–180 d; Group II: >180 d after transplantation).

From November 1998 to December 2007, fluconazole (400 mg/d, oral (p.o.)) was given to all HSCT recipients from Day −7 to Day +90 post-HSCT, with dosages adjusted on the basis of their renal function. After January 2008, patients with a history of IFI pre-HSCT received itraconazole (200 mg/d, intravenous (i.v.)) during neutropenia (<1.0×109 L−1), and were changed to oral fluconazole or itraconazole after neutrophil recovery (>1.0×109 L−1). Patients with persistent fever refractory to broad-spectrum antibiotic treatment were also provided with antifungal therapy, usually itraconazole, caspofungin, voriconazole, or amphotericin B.

2.6 Statistical methods

Continuous variables were compared between groups using Student’s t test or Wilcoxon rank sum test. The Chi-square test or Fisher’s test was used for categorical variables. Univariate analysis and multivariate analysis were performed, and Cox proportional hazard models were estimated to assess risk factors for IFI and survival rate. Kaplan-Meier estimates were computed for survival, and a stratified log-rank test was used to compare these groups. P<0.05 was considered significant. All analyses were conducted using SAS 9.2 and ATATA 11.0.

3 Results

3.1 Clinical outcomes after allo-HSCT

Clinical outcomes after allo-HSCT among 408 patients were analyzed and summarized in Table 2. The median follow-up period was 28 (range 1–145) months. All patients achieved hematopoietic reconstitution, with neutrophil engraftment (>0.5×109 L−1) on Day +14 (range +7–+31 d) and platelet engraftment (>20×109 L−1) on Day +18 (range +8–+144 d). The cumulative incidence rates of acute GVHD Grades I to II and Grades III to IV were 35.5% and 8.1%, respectively. The cumulative incidence of chronic GVHD was 39.2% (29.9% limited and 9.3% extensive). Positive CMV antigenemia was observed in 35.5% of the recipients. Thirty-nine patients developed CMV disease, including pneumonia, enteritis, and cystitis.

Table 2 Clinical outcomes after allo-HSCT *
Full size table

3.2 Incidence and clinical features of IFI

We detected 92 (22.5%) episodes of IFI in patients after allo-HSCT, with 4 proven cases and 88 probable IFI cases. The median age of these patients was 29 (range 9–50) years. Among the 92 cases, 69% were males. The cumulative incidence rates of IFI for 100 d, 6 months, and 1 year were 7.87%, 5.77%, and 12.60%, respectively (Fig. 1). Candidiasis was responsible for 50 proven and probable episodes (54.35%), and mold infection for 42 episodes (45.65%). Non-albicans caused 52% (26/50) of candidiasis. Among mold infection, Aspergillus was associated with 76.19% (32/42) cases (Fig. 2). The median time after allo-HSCT to onset of IFI was 140 (range 30–598) d for invasive candidiasis infection (ICI) and 243 (range 12–2461) d for invasive mold infection (IMI). There was a trend for ICI to occur earlier post-HSCT than IMI. The lung was the most frequently affected site (69.6%, 64/92), followed by oropharynx (14.1%, 13/92), gastrointestinal tract (9.8%, 9/92) and urinary tract (10.9%, 10/92), blood (3.3%, 3/92), and skin (1%, 1/92), and eight cases had more than one site affected.

Fig. 1
figure 1

Cumulative incidence rates of invasive fungal infection (IFI) among 408 patients after allo-HSCT

The cumulative incidence rates of IFI for 100 d, 6 months, and 1 year were 7.87%, 5.77%, and 12.60%, respectively

Full size image
Fig. 2
figure 2

Pathogen distribution of invasive fungal infection (IFI) in different time periods: early (≤100 d post-HSCT), late Group I (100□264;180 d post-HSCT), and late Group II (<180 d post-HSCT)

Full size image

According to the timing after allo-HSCT, 27 (29.35%) cases were diagnosed as early IFI (≤100 d) and 65 (70.65%) cases were diagnosed as late IFI (>100 d), of which 43 (66.15%) developed later than 180 d after transplantation. Candida accounted for 66.67% (18/27) and 49.23% (32/65) of early and late IFIs, respectively. However, this difference did not reach statistical significance. We further grouped late IFI to Group I (<+180 d) and Group II (>+180 d), revealing that the probability of Candida among Group I was significantly higher than that among Group II (72.72% vs. 37.21%; Fig. 2).

3.3 Risk factors for IFI

Risk factors by univariate analysis for the development of IFI during each time period are presented in Table 3. HLA mismatch, sustained neutropenia (<0.5×109 L−1) lasting >2 weeks, CMV infection, and severe acute GVHD were significant risk factors for early IFI. Chronic GVHD and CMV disease increased risk for late IFI. A prior history of IFI and corticosteroid therapy associated with increased risk for both early and late IFIs. Age, gender, underlying disease, and conditioning regimen were not significant.

Table 3 Univariate analysis of risk factors for early IFI (≤100 d post-HSCT) and late IFI (>100 d post-HSCT)
Full size table

Multivariate analyses showed previous history of IFI (hazard ratio (HR)=9.53; 95% confidence interval (CI), 4.05–22.44), HLA mismatch (HR=2.55; 95% CI, 1.07–6.06), sustained neutrophil >2 weeks (HR=1.94; 95% CI, 1.03–4.56), and Grades III to IV acute GVHD (HR=2.46; 95% CI, 1.01–12.46) were significant for early IFI. A prior history of IFI (HR=2.62; 95% CI, 1.46–4.72), CMV disease (HR=3.80; 95% CI, 1.71–8.42), limited and extensive chronic GVHD (HR=2.84; 95% CI, 1.52–5.30 and HR=3.21; 95% CI, 1.54–6.68, respectively) demonstrated a higher mortality risk for late IFI. The impact of corticosteroids for late IFI depended on the dosage. Corticosteroids of <1.0 mg/(kg·d) did not increase risk of developing IFI. Conversely, corticosteroids with a dosage of ≥1.0 mg/(kg·d) affected IFI in the late period (Table 4).

Table 4 Multivariate analysis of risk factors for early and late IFIs
Full size table

3.4 Outcomes

IFI-related mortality was 49 out of 92 cases (53.26%) of IFI. The 12-year overall survival (OS) rate for IFI was 41.9%, with significant differences when compared with patients without IFI (63.6%, P<0.001; Fig. 3a). Late IFI was associated with a significantly higher 12-year OS compared with early IFI (Fig. 3b). For patients with only one IFI, the 12-year OS was better for patients with Candida when compared with molds (Fig. 3c).

Fig. 3
figure 3

Twelve-year overall survival of invasive fungal infection

(a) The overall survival for all 408 patients with or without IFI; (b) Overall survival for 27 patients with development of early IFI and 65 patients with late IFI; (c) Overall survival for patients with Candida infection and mold infection

Full size image

4 Discussion

Despite considerable progress in the management of the complications of HSCT, infection remains an important cause of post-transplant morbidity and mortality, primarily after allogeneic HSCT. European Group for Blood and Marrow Transplantation (EBMT) analyzed a large homogeneous group of 14 403 patients transplanted for early leukemia from an HLA-identical sibling and cause of infection death after allogeneic HSCT (Gratwohl et al., 2005). Survival increased from 52% at 5 years in the first to 62% in the third cohort (P<0.05), and treatment-related mortality decreased from 36% to 26% (P<0.05) due to a reduction in death from infection (P<0.01).

IFI is an important cause of post-transplant mortality. The incidence rate of IFI has been reported to be 10%–26% in HSCT recipients (Koldehoff and Zakrzewski, 2005; Post et al., 2007), with mortality ranging from 40% to 90% (Lin et al., 2001; Dagenais and Keller, 2009).

We performed this 12-year retrospective study to evaluate the epidemiologic characteristics, clinical manifestation and outcomes of IFI in 408 allogenic HSCT recipients. We report that patients with IFI have a lower 12-year OS than those without IFI (41.9% vs. 63.6%, P<0.01), reflecting the fact that IFI has a significant impact on long-term survival of transplant patients.

Over the past two decades, changes in transplantation practices and strategies to diagnose and treat IFI have likely impacted the epidemiology of IFI. According to the data from the Transplant Associated Infections Surveillance Network, a network of 23 US transplant centers, invasive aspergillosis (43%), invasive candidiasis (28%), and zygomycosis (8%) are the most common IFIs. Aspergillus fumigatus causes 44% of aspergillosis, and Candida glabrata is the most common organism (33%) causing invasive candidiasis (Kontoyiannis et al., 2010). In our previous study, we performed CT-guided percutaneous lung biopsy in 16 patients who were initially suspected of developing fungal infection. Ten out of the 16 patients (62.5%) were diagnosed with a fungal infection (8 with Aspergillus, 2 with mold fungus). All of the 10 cases had mold infection, and the incidence of Aspergillus infection was 80% (Shi et al., 2009).

The observed results in our study are not entirely consistent with the prior studies. According to our data, candidiasis was more prevalent than mold infections (54.35% vs. 45.65%). Among the candidiasis cases, Candida albicans was the most common pathogen (28%), followed by Candida glabrata (18%) and Candida tropicalis (14%). The rate of mold infections was lower than that reported in our previous study (Shi et al., 2009), which may be explained by the following reasons. All patients underwent CT-guided lung biopsies in our previous prospective study, providing higher diagnostic accuracy than blood and sputum culture, while only 4 patients underwent biopsies in this retrospective study. Also, patients who received lung biopsies were highly suspected of Aspergillus infection based on imaging findings, probably leading to an increased rate of Aspergillus infection. Additionally, the small number of patients (16 cases) may also amplify the positive rate of mold infection.

Although Candida was presented as the most common pathogen in our center, there seemed to be a trend toward higher numbers of non-albicans and molds due to routine prophylaxis with fluconazole. This serves as a reminder to choose broad-spectrum antibiotics covering non-albicans and molds in the treatment of post-HSCT fungal infections. Since only a small number of patients were treated with voriconazole in our study, we observed no voriconazole-resistant species (e.g. Zygomycetes).

Previous studies have noted that IFI primarily occurs in the phase of hematopoietic reconstitution, generally two to three months after transplantation. Candida is the most common pathogen of early IFI, while mold is the most common pathogen of late IFI. In patients who also developed GVHD, IFI may occur at three to six months and even later (Koldehoff and Zakrzewski, 2005; Garcia-Vidal et al., 2008). Our data demonstrate that the median time for the development of ICI is earlier than that for IMI (140 d vs. 243 d, P<0.05), reflecting the fact that mold infections are more common in the late period after HSCT. The incidences of mold and Candida infections in early and late IFIs differed, but did not reach statistical significance. We further evaluated the incidence rate of very late IFI and found that mold infections significantly increased (P<0.05). These findings highlight the importance of strengthening the detection and treatment of mold in the late period after transplantation.

Multiple factors reported to be associated with the risk of IFI include donor type, conditioning regimens, prolonged neutropenia, GVHD, and high-dose corticosteroids (Fukuda et al., 2003; Thursky et al., 2004; Ozyilmaz et al., 2010; Omer et al., 2013). According to our data, we found similar risk factors infecting IFI. However, when we further analyzed the risk factors for each phase of IFI, we were able to demonstrate that there is a difference between the early and late IFIs.

Despite the widespread use of cyclosporine, tacrolimus, mycophenolate mofetil, and other drugs for prophylaxis of GVHD in HSCT recipients, 25% to 45% of patients still developed Grades II to IV acute GVHD. HLA-mismatched recipients are at a particularly high risk for acute GVHD, which is the most likely cause of the high risk for IFI, especially for early IFI. Enhancing the immune inhibitors or increasing the dosage of corticosteroids may effectively control GVHD, while increasing the risk of IFI.

In our study, univariate analysis showed that corticosteroid therapy is a risk factor during all periods. When further introduced into multivariate analyses, corticosteroid therapy only accounted for late IFI. This result suggests that long-term use of corticosteroids has a more advanced effect on late IFI.

CMV is associated with development of IFI in both the early and the late post-transplant phases (Marr et al., 2002). The mechanism remains unclear. The immune-modulating effect of CMV may be one of the explanations. In our study, CMV infection was the risk factor for early IFI in univariate analysis but not in multivariate analyses. CMV disease rather than CMV infection was associated with the risk of late IFI. This may be because CMV mostly occurs in the late period after transplantation. Strengthening serum monitoring of CMV and timely prevention of CMV may reduce the incidence of IFI, particularly late IFI.

Prolonged neutropenia is a risk factor for IFI (Hovi et al., 2000). Neutrophils play an important role in the defense against IFI. Our results showed that prolonged neutropenia, which generally occurs in the early period after transplantation, is a risk factor for early IFI but not for late IFI.

It is important to recognize that the risks for IFI also depend on changes in the conditioning regimens, supportive care strategies, and therapy strategies. This study attempts to analyze the impact of conditioning regimens of IFI, such as myeloablative and non-myeloablative regimens, as well as conditioning regimens with or without ATG. Unfortunately we cannot draw convincing conclusions due to the small number of cases. Further work will be needed to assess the association between conditioning regimen and IFI.

It should be noted that our study has some limitations. Firstly, incidence of IFI can be lower than the actual incidence because of the strict definition of IFI (our study did not include possible IFI cases). Secondly, some patients with IFI may be misplaced into the control group, thus the significance of the difference between the two groups may be diluted. What is more, 125 of the total 408 patients participated in clinical trials of secondary prophylaxis during the study period, and patients with a history of IFI pre-HSCT received itraconazole (200 mg/d, i.v.) during neutropenia (<1.0×109 L−1), and changed to oral fluconazole or itraconazole after neutrophil recovery (>1.0×109 L−1), which in part may reduce the incidence of IFI after HSCT.

Compliance with ethics guidelines

Ji-min SHI, Xu-ying PEI, Yi LUO, Ya-min TAN, Ru-xiu TIE, Jing-song HE, Wei-yan ZHENG, Jie ZHANG, Zhen CAI, Mao-fang LIN, and He HUANG declare that they have no conflict of interest.

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2008 (5). Informed consent was obtained from all patients for being included in the study. Additional informed consent was obtained from all patients for which identifying information is included in this article.