ABSTRACT
Limited data are available on micafungin breakthrough fungemia (MBF), fungemia that develops on administration of micafungin, in patients with hematological disorders. We reviewed medical and microbiological records of patients with hematological disorders who developed MBF between January 2008 and June 2015. A total of 39 patients with MBF were identified, and Candida (30 strains) and non-Candida (9 strains) fungal species were recognized as causative strains. Among 35 stored strains, Candida parapsilosis (14 strains), Trichosporon asahii (7 strains), Candida glabrata (5 strains), and other fungal species (9 strains) were identified by sequencing. Neutropenia was identified as an independent predictor of non-Candida fungemia (P = 0.023). T. asahii was the most common causative strain (7/19) during neutropenia. The 14-day crude mortality rate of patients treated with early micafungin change (EMC) to other antifungal agents was lower than that of the patients not treated with EMC (14% versus 43%, P = 0.044). Most of the stored causative Candida strains were susceptible (80%) or showed wild-type susceptibility (72%) to micafungin. The MICs of voriconazole for T. asahii were low (range, 0.015 to 0.12 μg/ml), whereas the MICs of amphotericin B for T. asahii were high (range, 2 to 4 μg/ml). MBF caused by non-Candida fungus should be considered, especially in patients with neutropenia. EMC could improve early mortality. Based on epidemiology and drug susceptibility profiling, empirical voriconazole-containing therapy might be suitable for treating MBF during neutropenia to cover for T. asahii.
INTRODUCTION
Fungemia, mostly due to Candida spp., rarely occurs in patients with cancer and hematological malignancies but is associated with substantial mortality (1). Recently, breakthrough fungemia, which develops upon administration of antifungal agents (AFAs), has been increasingly recognized in patients with hematological disorders (2–4). Previous studies reported that the distribution of causative fungal species depends on the AFA administered when breakthrough fungemia develops (2). However, studies of breakthrough fungemia during administration of echinocandins, including micafungin, are rare, although echinocandin prophylaxis and empirical echinocandin administration have been approved as current clinical practices for high-risk patients with hematological diseases (5–8). Thus, we conducted this study, which focused on micafungin breakthrough fungemia (MBF), to clarify the characteristic features and determine the appropriate therapeutic strategy.
(This work was presented in part at ID Week 2017, San Diego, CA, 4 to 8 October 2017.)
RESULTS
Clinical presentation and characteristics of patients with MBF.A total of 39 patients with MBF were identified during the study period. Of the 39 patients, 36 received ≥150 mg/day of micafungin when MBF developed. Clinical characteristics of patients with MBF are shown in Table 1. The median duration of micafungin exposure prior to breakthrough was 40 days (range, 4 to 137 days). The most common underlying hematological disorder was acute myeloid leukemia. Nineteen patients developed MBF in the presence of neutropenia. Additionally, 9 patients developed catheter-related fungemia, and all had candidemia. Among the 34 patients who were treated with an empirical AFA, 27 patients received liposomal amphotericin B (69%); 14-day and 30-day crude mortality rates were 33% and 46%, respectively.
Clinical characteristics of 39 patients with MBF
Causative fungal species of MBF.Among the 39 causative fungal strains, 9 were non-Candida strains. The 35 causative fungal strains that were stored and identified by sequencing were Candida parapsilosis (14 strains), Trichosporon asahii (7 strains), Candida glabrata (5 strains), Candida albicans (3 strains), Candida guilliermondii (2 strains), Candida krusei (1 strain), Candida fermentati (1 strain), Cryptococcus neoformans (1 strain), and Fusarium dimerum species complex (1 strain). The remaining 4 nonstored strains were identified only using Vitek or VITEK2 as C. parapsilosis (2 strains), C. krusei (1 strain), and Candida tropicalis (1 strain).
Differences between candidemia and non-Candida fungemia.Univariate analysis revealed that the proportion of patients who developed non-Candida fungemia in the presence of neutropenia was higher than that of patients who developed candidemia in the presence of neutropenia (P = 0.008) (Table 2). Additionally, neutropenia was identified as an independent predictor of non-Candida fungemia among MBF patients in multivariate analysis (adjusted odds ratio, 13.8; 95% confidence interval, 1.44 to 132).
Differences between patients with candidemia and patients with fungemia other than candidemiaa
Distribution of causative fungal species in the presence of neutropenia (absolute neutrophil count [ANC] ≤ 500/μl) and nonneutropenia (ANC > 500/μl).T. asahii was the most common causative strain (37%) in patients with neutropenia (Table 3). However, 19 of the 20 causative fungal strains (95%) in nonneutropenic patients were Candida strains (95%).
Distribution of causative MBF species during neutropenia and nonneutropeniaa
Microbiological characteristics of MBF.The microbiological characteristics of the 35 fungal strains stored for sequencing-based identification and antifungal drug susceptibility were determined. The remaining 4 nonstored strains were excluded.
(i) Non-Candida strains.Nine patients had fungemia caused by non-Candida fungal strains. All 9 causative fungal strains were stored and analyzed (Table 4, patients 1 to 9) and were properly identified by using matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS). All 9 strains had a micafungin MIC of >16 μg/ml. MICs of voriconazole and amphotericin B for all 7 T. asahii strains were 0.015 to 0.12 μg/ml and 2 to 4 μg/ml, respectively. Other susceptibility results are shown in Table 4.
Microbiological characteristics of MBFa
(ii) Candida strains.In total, 26 of the 30 Candida strains that caused MBF were stored and analyzed (Table 4), and 24 of the 26 (92%) were accurately identified with MALDI-TOF MS. However, C. fermentati was not identified, and a C. parapsilosis strain was misidentified as Candida metapsilosis by MALDI-TOF MS. The micafungin MICs for all 26 strains ranged from 0.015 to 4 μg/ml. C. fermentati has no established breakpoint, and no epidemiological cutoff value (ECV) for C. fermentati has been established for micafungin. Among the 25 strains for which breakpoints and an ECV were established for micafungin, 18 strains showed wild-type susceptibility to micafungin and 20 strains were susceptible to micafungin. Nevertheless, C. parapsilosis was significantly more frequent in micafungin breakthrough candidemia (MBC) caused by micafungin-susceptible Candida strains (14/20) than in MBC caused by micafungin-nonsusceptible Candida strains (0/5) (P = 0.0087) (see Table S1 in the supplemental material). All 14 C. parapsilosis strains were not resistant to fluconazole and voriconazole (Table 4); no C. glabrata strain was susceptible to fluconazole. Also, 4 of the 5 C. glabrata strains showed non-wild-type susceptibility to voriconazole.
Analysis of risk factors for crude 30-day mortality of MBF.In univariate analysis, an age of ≥60 years, chronic renal failure, and septic shock were significant risk factors for 30-day crude mortality (Table 5). Additionally, mortality in patients with Trichosporon fungemia tended to be higher than that in patients with candidemia (71% versus 41%, P = 0.08).
Risk factor analysis for crude 30-day mortality of MBFa
In multivariate analysis, an age of ≥60 years, chronic renal failure, septic shock, and systemic steroid administration were identified as independent risk factors for mortality (Table 5).
Efficacy of early empirical micafungin change to other antifungal agents.Among the 35 patients who survived for 48 h after the onset of MBF, 21 patients received early micafungin change to other AFA (EMC) treatment. The 14-day crude mortality rate of the EMC group was significantly lower than that of the non-EMC group (Fig. 1) (14% versus 43%, P = 0.044), but baseline characteristics of the two groups did not differ significantly (see Table S2 in the supplemental material). Also, the 30-day crude mortality rate of the EMC group was lower than that of the non-EMC group, but the difference was not significant (33% versus 50%, P = 0.21).
Fourteen-day crude mortality rates of the EMC group and the non-EMC group. EMC, early micafungin change to other antifungal agents.
DISCUSSION
To the best of our knowledge, this is the first and largest study clarifying the characteristic features of MBF and determining the appropriate therapeutic strategy. The administration of echinocandins, including micafungin, is recommended as empirical and prophylactic AFA administration in some high-risk patients (5–7, 9). However, echinocandin breakthrough fungemia, including MBF, is recognized as a severe infection among patients administered echinocandins (10–13). Thus, it is important to establish the appropriate therapeutic strategy for MBF.
The exact incidence of MBF among patients who receive micafungin administration remains unknown. However, based on our previous study, it is predicted to be low, because only 17 cases with MBC were documented among 768 cases who underwent allogeneic hematopoietic stem cell transplantation (allo-HSCT) (14). In the present study, most of the Candida species that caused MBC were susceptible or showed wild-type susceptibility to micafungin (Table 4). These results were the same as those of our previous study (14). Furthermore, C. parapsilosis was significantly more frequent in MBC caused by micafungin-susceptible Candida strains than in MBC caused by micafungin-nonsusceptible Candida strains (P = 0.0087) (see Table S1 in the supplemental material); C. parapsilosis was the most common causative Candida species in MBC caused by micafungin-susceptible Candida strains (14/20). These results indicate that the current breakpoint of micafungin for C. parapsilosis established by the Clinical Laboratory and Standards Institute (CLSI) may not be well correlated with prophylactic efficacy of micafungin for C. parapsilosis in patients with high-risk hematological disorders.
In this study, EMC significantly improved the 14-day crude mortality rate (Fig. 1). Among all 39 patients, EMC significantly improved 14-day crude mortality (adjusted hazard ratio, 0.089; 95% confidence interval, 0.0083 to 0.95). A clear understanding of the epidemiology of MBF is essential to determine the appropriate empirical AFA. Table 3 shows the causative fungal strains of MBF. Similar fungal strains, except for the Aspergillus and Mucor species that are rarely cultured from blood samples, were reported to be the causative strains of echinocandin breakthrough invasive fungal infections in some previous reports (11–13). Thus, our data are likely applicable to MBF in other clinical settings. In our study, neutropenia was an independent predictor of fungemia caused by non-Candida fungal strains (Table 2). Neutropenia was also a predictor of T. asahii fungemia in univariate analysis (P = 0.003), and T. asahii was the most common strain causing MBF in patients who developed neutropenia (Table 3). Identifying this kind of predictor is important because most non-Candida fungal species cannot be distinguished from Candida species merely by microscopic appearance after Gram staining of positive blood cultures. A systematic review revealed that 84.8% (67/79) of Trichosporon infections among patients with hematological disorders occurred in those with neutropenia (15). Accordingly, the therapeutic range that covers T. asahii is reasonable in empirical treatment of MBF in patients who develop neutropenia. Current guidelines recommend voriconazole as the preferred therapeutic AFA for treating invasive Trichosporon infection, including fungemia (16, 17), but no randomized controlled trial (RCT) has been conducted to compare voriconazole and other AFAs, including liposomal amphotericin B, for Trichosporon infection. Furthermore, in the above-mentioned systematic review, among patients with Trichosporon infection and hematological disorders, voriconazole-based therapy was associated with more favorable outcomes than therapy with other AFAs (73.6% versus 41.8%, P = 0.016) (15). In our study (Table 4) and a previous study (18), the MICs of voriconazole for Trichosporon were low, whereas the MICs of amphotericin B for Trichosporon were high. Although the MICs of amphotericin B were ≥2 μg/ml (27%) for only 7 of the 26 Candida strains tested for susceptibility, the MICs were ≥2 μg/ml for the 7 T. asahii strains in our study (100%) (P < 0.001). Also, only 1 patient among those who developed T. asahii fungemia received empirical voriconazole treatment (Table 4). That might account for the high 30-day crude mortality rate of patients with T. asahii fungemia (71%). In addition, voriconazole could typically cover C. parapsilosis, which was the second most common causative fungal strain in patients who developed neutropenia (Table 3). Hence, voriconazole might be a suitable appropriate empirical AFA for MBF in patients with neutropenia.
The MICs of azoles, including voriconazole, for C. glabrata (one of the causative fungal strains of MBF) are known to be high (9), and some of the causative fungal strains had high voriconazole MICs (Table 4). Accordingly, combining liposomal amphotericin B and voriconazole might be a more appropriate empirical therapy, especially for neutropenic patients with MBF who also have risk factors for 30-day crude mortality (Table 5). Liposomal amphotericin B might be a reasonable AFA for MBF, especially in nonneutropenic patients, because almost all the causative fungi in nonneutropenic patients were Candida species, which showed wild-type susceptibility to liposomal amphotericin B (Table 4). Furthermore, 16 of the 21 patients in the EMC group received liposomal amphotericin B empirically and EMC improved early mortality significantly (Fig. 1). In our study, empirical voriconazole administration might be difficult in some patients, such as allo-HSCT recipients, because of the possible interactions with other drugs, such as tacrolimus, that are often administered in such patients (19).
This study demonstrated the utility of MALDI-TOF MS for identifying the causative fungal strains of MBF (Table 4). The accuracy of MALDI-TOF MS-based identification in this study is comparable to that in previous reports (20, 21). Thus, MALDI-TOF MS might be useful for providing rapid detection and identification of causative fungal strains of MBF.
Our study has some limitations. First, this was a small retrospective study, not an RCT, conducted among patients with MBF or Trichosporon infection to show the superiority of voriconazole over liposomal amphotericin B or other AFAs. However, conducting RCTs is expected to be challenging. Therefore, larger retrospective studies are needed to determine whether the results of our study are applicable in other clinical settings. Second, AFA susceptibility testing and sequencing-based identification were not performed on all 39 fungal strains, because 4 of the 30 Candida strains were lost. Third, EMC did not improve the 30-day crude mortality rate significantly, although the mortality rate in the EMC group was lower than that of the non-EMC group. The number of patients with MBF was probably insufficient to show the statistical significance. Nevertheless, early initiation of AFA has been reported to improve mortality in patients with invasive fungal infections (22, 23), including breakthrough candidemia (23). Therefore, the efficacy of EMC in reducing the mortality rate is highly considerable. Fourth, Trichosporon might be more common in our institute than in other settings. Recently, cases of Trichosporon fungemia have been reported mainly from countries in Asia, including Japan (24). However, the exact epidemiology of Trichosporon infection in Asia and other regions is yet unknown.
In summary, MBF caused by non-Candida fungal strains should be considered, especially in patients who develop neutropenia. EMC could improve early mortality in patients with MBF. Based on the epidemiology and drug susceptibility profile of MBF, early empirical voriconazole-containing therapy might be suitable for treating MBF in patients with neutropenia, with additional therapeutic coverage for T. asahii.
MATERIALS AND METHODS
We conducted a retrospective analysis of MBF among patients (age ≥ 20 years) with hematological disorders between January 2008 and June 2015 at Toranomon Hospital (Tokyo, Japan; 890 beds) and Toranomon Hospital Kajigaya (Kanagawa, Japan; 300 beds). The medical and microbiological records of these patients were reviewed. This study was approved by the institutional review board (IRB) of Toranomon Hospital. Although the recommended prophylactic dose of micafungin is 50 mg/day (6, 7), a dose of 150 mg/day was administered to some high-risk patients as antimold prophylaxis, in accordance with a clinical trial also approved by the IRB (14).
Definitions.Fungemia was defined as the isolation of any fungal species in ≥1 set of blood culture samples obtained from patients with clinical features. MBF was defined as fungemia occurring in patients receiving intravenous micafungin (≥50 mg/day) for ≥3 days before the first positive blood culture was obtained. This definition is the same as that of breakthrough candidemia (14, 25). A second episode of fungemia occurring in the same patient within 4 weeks of the first episode was considered the same episode (26). Neutropenia was defined as an ANC of ≤500 cells/μl. Administration of systemic steroids or immunosuppressive agents was defined as usage within a period of 90 days before the onset of fungemia. Chronic renal failure was defined as an estimated glomerular filtration rate of <60 ml/min/1.73 m2 over 12 weeks before the onset of fungemia. Septic shock was defined as persistent sepsis-induced hypotension despite adequate resuscitation or cases requiring vasopressor agents (27). Catheter-related fungemia was diagnosed when there was no other apparent source of infection and the fungal strain was isolated from both peripheral blood and catheter tip cultures (14). Empirical AFA was defined as an AFA that was started to treat MBF immediately after discontinuation of micafungin administration. Early micafungin change to other AFAs (EMC) meant that micafungin was switched to empirical AFAs within 48 h after the onset of MBF.
Identification.Blood culture samples were processed using Bactec 9240 (between January 2008 and June 2015) and Bactec FX (between June 2010 and June 2015) systems (Becton Dickinson and Company, Sparks, MD, USA). All breakthrough fungal isolates were recovered on Sabouraud dextrose agar (Nippon Becton Dickinson and Company, Ltd., Japan) at 35°C and identified to the species level by using the Vitek or VITEK2 systems (bioMérieux, Marcy l'Etoile, France) for all fungal strains in the Toranomon Hospital. The internal transcribed spacer (ITS) regions and D1/D2 regions of the rRNA gene of the isolates were sequenced to provide further supportive evidence (28, 29). In addition to these regions, the elongation factor gene (EF-1α) was sequenced for discrimination of Fusarium species (30, 31). The stored strains were also analyzed and identified using the matrix-assisted laser desorption ionization-time of flight mass spectrometry Biotyper preprocessing standard method 1.1 (Bruker Daltonics, Billerica, MA, USA).
Antifungal susceptibility.In vitro susceptibilities of the stored fungal strains obtained from the blood samples to five AFAs (fluconazole, itraconazole, voriconazole, amphotericin B, and micafungin) were determined using the broth microdilution method with a commercial frozen plate for antifungal susceptibility testing (Eiken Chemical Co., Ltd.) that complied with CLSI guideline M27-A3 (32), in accordance with the manufacturer's instructions, and MIC values were interpreted according to the criteria in CLSI testing guideline M27-S4 for yeasts (33), M38-A2 for molds (34), and the ECV for Candida strains (35). C. parapsilosis ATCC 22019 and C. krusei ATCC 6258 were used as quality control isolates.
Evaluating the efficacy of early empirical micafungin change to other antifungal agents.To evaluate the efficacy of EMC in reducing mortality, patients who were treated with EMC (EMC group) were compared with those who were not treated with EMC (non-EMC group). Patients who died within 48 h after the onset of MBF were excluded from this analysis, ensuring that the potential impact of early (≤48-h) micafungin change was properly evaluated. A similar method was used in a previous study on candidemia, which evaluated the outcome of the therapeutic strategy (36). We selected 14-day crude mortality as the primary endpoint, because it was thought to be most reflective of death attributable to MBF. The second outcome was 30-day crude mortality. The same endpoints were used as in the previous study on bloodstream infection for a similar reason (37).
Statistical analysis.Categorical variables were compared using Fisher's exact test. Variables with P values of <0.20 were included in a logistic regression model for multivariate analysis. The 30-day mortality rates were estimated using Kaplan-Meier analysis. Risk factors associated with crude 30-day mortality were analyzed using the log rank test for pre-MBF onset variables with a Cox proportional-hazard regression model for a time-dependent variable (early central venous catheter removal) for univariate analysis. Variables with P values of <0.10 were included in a Cox proportional-hazard regression model with a stepwise method for multivariate analysis. Significance was set at an α of 0.05. All statistical analysis was performed with EZR (Saitama Medical Center, Jichi Medical University, Japan), a graphical user interface for R (The R Foundation for Statistical Computing) (38).
ACKNOWLEDGMENTS
This study was supported by the Research Program on Emerging and Re-emerging Infectious Diseases of the Japan Agency for Medical Research and Development (AMED) under grant number JP17fk0108208.
Identification and drug susceptibility testing was performed by the staff of the microbiology laboratory of Toranomon Hospital and the staff of the Department of Chemotherapy and Mycoses, National Institute of Infectious Diseases.
We report no conflicts of interest. We alone are responsible for the content and writing of this paper.
FOOTNOTES
- Received 24 October 2017.
- Returned for modification 21 December 2017.
- Accepted 25 February 2018.
- Accepted manuscript posted online 12 March 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/AAC.02183-17.
- Copyright © 2018 American Society for Microbiology.
