Invasive Non-Aspergillus Mold Infections in Transplant Recipients, United States, 2001–2006

Recent reports describe increasing incidence of non-Aspergillus mold infections in hematopoietic cell transplant (HCT) and solid organ transplant (SOT) recipients. To investigate the epidemiology of infections with Mucorales, Fusarium spp., and Scedosporium spp. molds, we analyzed data from the Transplant-Associated Infection Surveillance Network, 23 transplant centers that conducted prospective surveillance for invasive fungal infections during 2001-2006. We identified 169 infections (105 Mucorales, 37 Fusarium spp., and 27 Scedosporium spp.) in 169 patients; 124 (73.4%) were in HCT recipients, and 45 (26.6%) were in SOT recipients. The crude 90-day mortality rate was 56.6%. The 12-month mucormycosis cumulative incidence was 0.29% for HCT and 0.07% for SOT. Mucormycosis incidence among HCT recipients varied widely, from 0.08% to 0.69%, with higher incidence in cohorts receiving transplants during 2003 and 2004. Non-Aspergillus mold infections continue to be associated with high mortality rates. The incidence of mucormycosis in HCT recipients increased substantially during the surveillance period.

Recent reports describe increasing incidence of non-Aspergillus mold infections in hematopoietic cell transplant (HCT) and solid organ transplant (SOT) recipients. To investigate the epidemiology of infections with Mucorales, Fusarium spp., and Scedosporium spp. molds, we analyzed data from the Transplant-Associated Infection Surveillance Network, 23 transplant centers that conducted prospective surveillance for invasive fungal infections during [2001][2002][2003][2004][2005][2006]. We identifi ed 169 infections (105 Mucorales, 37 Fusarium spp., and 27 Scedosporium spp.) in 169 patients; 124 (73.4%) were in HCT recipients, and 45 (26.6%) were in SOT recipients. The crude 90-day mortality rate was 56.6%. The 12-month mucormycosis cumulative incidence was 0.29% for HCT and 0.07% for SOT. Mucormycosis incidence among HCT recipients varied widely, from 0.08% to 0.69%, with higher incidence in cohorts receiving transplants during 2003 and 2004. Non-Aspergillus mold infections continue to be associated with high mortality rates. The incidence of mucormycosis in HCT recipients increased substantially during the surveillance period.
I nvasive mold infections are a major source of illness and death in transplant recipients. Non-Aspergillus invasive infections are of particular concern because of the following factors: the diffi culty in distinguishing them clinically from Aspergillus spp. infections and from each other; their progressive and aggressive course; and the intrinsic resistance of many of these fungi to several antimicrobial agents, including voriconazole. Recent reports have described an increase at some medical centers in non-Aspergillus mold infections, particularly mucormycosis (formerly zygomycosis), often in persons receiving antifungal agents that have activity against Aspergillus spp. (1)(2)(3)(4).
Most reports have come from single institutions and might not be broadly representative of general trends. A comprehensive look at the modern epidemiology of these infections, including trends at multiple sites, can increase understanding of their public health implications. The Transplant-Associated Infection Surveillance Network (TRANSNET), a multicenter network of 23 academic tertiary care medical centers in the United States, performed prospective surveillance for invasive fungal infections (IFIs) during 2001-2006 (5-7). We analyzed the uniquely comprehensive TRANSNET cohort for epidemiology, patient demographics, clinical features, and outcomes of infections caused by the most common non-Aspergillus invasive molds detected: those of the order Mucorales, Fusarium spp., and Scedosporium spp.

Methods
Study personnel based at each TRANSNET center reviewed records of patients who had undergone solid organ transplantation (SOT) or hematopoietic cell transplantation (HCT). Each IFI was evaluated to determine whether it met the criteria for a proven or probable IFI as defi ned by the European Organization for Research and Treatment of Cancer and Mycoses Study Group (EORTC-MSG) (8) Cultures and histopathologic specimens were processed at the participating hospitals. Species were identifi ed by using routine methods at the local laboratories. Fungal isolates were forwarded to the University of Alabama at Birmingham Fungal Reference Laboratory (Birmingham, AL, USA) and the Mycotic Diseases Branch at the Centers for Disease Control and Prevention (Atlanta, GA, USA), where species identifi cation was confi rmed by using morphologic and DNA-based methods. Patients in this analysis had an IFI and 1) a culture or histopathologic specimen consistent with Mucorales mold or 2) a culture from a clinical specimen positive for Fusarium spp. or Scedosporium spp. We excluded patients with invasive mold infection for whom diagnostic evidence was inconclusive.
All detected cases, regardless of transplant date, were described (surveillance cohort). In addition, the subset of cases in persons receiving transplants during the surveillance period was included for determination of the incidence of these infections (incidence cohort). Denominator data collected on identifi ed patients undergoing transplant at each site during the surveillance period included date and type of transplant and date of last follow-up and clinical status at that time (alive, had undergone retransplant, IFI, dead). These data were used to calculate transplant-associated 12-month cumulative incidence. Estimated cumulative incidence was based on time to fi rst infection after transplant and accounted for the competing risks for death, relapse of underlying disease, and retransplant. For HCT recipients, overall and transplant-specifi c 12-month cumulative incidence was calculated for mucormycosis and for fusariosis and scedosporiosis combined. For SOT patients, overall and transplant-specifi c incidence was estimated only for mucormycosis because of low numbers of scedosporiosis and fusariosis. We estimated cumulative incidence using the cmprisk risk package version 2.2-1 in R version 2.11.1 (www.r-project.org).
We also explored the change in incidence of mold infections over time among HCT recipients. Because case counts can vary due to increases or decreases in the size of the denominator, we measured the changing epidemiology by estimating the incidence, according to transplant date. To do this, we classifi ed the denominator of all transplant patients into 9 sequential subcohorts according to the 4-month period when the transplant was performed, as described (5,7). Twelve-month cumulative incidence for each of the 9 time periods (i.e., subcohorts) was calculated for mucormycosis and for fusariosis and scedosporiosis combined.
In addition, we attempted to estimate whether the underlying risk for infection in the overall population was changing because of changes in the numbers of allogeneic HCTs. To do this, we calculated the total follow-up time (in person-days) for each transplant type and plotted it for each period. Analysis was limited to the 21 sites that contributed data consistently during May 2002-April 2005, the last period for which complete data were available.

Clinical Description of Cases
Of the 1,208 cases of proven or probable IFI in SOT recipients (7) and the 983 cases in HCT recipients (5) in TRANSNET, we detected 169 cases of Mucorales, Fusarium spp., or Scedosporium spp. infection in 166 transplant recipients, making these molds the most frequently identifi ed molds, after Aspergillus within this patient population. The Mucorales (105 patients) were the most common of these molds, followed by Fusarium spp. (37 patients), and Scedosporium spp. (27 patients ranging from 89 to 2,551 HCTs and from 239 to 2,111 SOTs ( Table 2). The median number of cases at a site was 8 (interquartile range 2-11 cases).
Of the 45 transplant patients with mold infections who had received an SOT, 19 (42.2%) had received a lung and 12 (26.7%) had received a kidney (  recipients was 81 days, compared with 533 for non-liver SOT recipients (p = 0.035).

Incidence Cohort
Follow-up data were available for 15,820 HCT recipients from 22 sites. Among these, 17% died, 13% experienced relapse, and 1% underwent retransplant in the fi rst 12 months after transplant. Mucormycosis was diagnosed in 44 (0.3%) persons. Overall 12-month cumulative incidence for mucormycosis in HCT recipients was 0.29% ( Figure 2). Recipients of an organ or cells from an allogeneic HLA-unrelated donor had the highest 12-month cumulative incidence of mucormycosis at 0.85%; allogeneic HLA-matched related donors had a cumulative incidence of 0.58%; and allogeneic HLA-mismatched related donors had a 12-month cumulative incidence of 0.25% (Figure 2).
Follow-up data were available for 16,457 SOT recipients from 15 sites. Of these, 6% died; 2% of kidney transplant recipients returned to chronic dialysis; and 2% underwent retransplant within 12 months after the transplant. In 12 (0.07%) persons, mucormycosis was diagnosed within the fi rst 12 months after transplant. In SOT recipients, the overall 12-month cumulative incidence of mucormycosis was 0.07% ( Figure 3). Patients receiving lung transplants and liver transplants had the highest incidence of mucormycosis (0.18% and 0.16%,  We estimated cumulative incidence for mucormycosis and for Fusarium spp. and Scedosporium spp. infections combined for the 9 subcohorts from May-August 2002 through January-April 2005 (Figure 4). Cohorts ranged from ≈1,200 to 1,400 transplant recipients each. Estimated cumulative incidence of Fusarium spp. and Scedosporium spp. infections in the 9 subcohorts did not increase over time. Incidence ranged from 0% during January-April 2004 and September-December 2004 to a high of 0.30% (4 infections) during May-August 2004.
The incidence of mucormycosis in HCT recipients varied, with generally higher incidences in 2003 and 2004 than in 2002. Initially, the 12-month cumulative incidence for mucormycosis was 0.08% (1 infection) for the cohort receiving transplants during May-August 2002. The cumulative incidence was higher for all subsequent cohorts and peaked at 0.69% (9 infections) in the cohort receiving transplants during September-December 2004. In the subsequent cohort that received transplants during January-April 2005, incidence declined to 0.16% (2 infections). These 39 Mucorales infections occurred at 13 of the 21 sites, with these 13 sites contributing 1-7 infections each. The pattern was similar across the 9 subcohorts when we excluded the site with most cases (site D) and limited the analysis to allogeneic transplants only (data not shown). In addition, the total follow-up time of allogeneic HCTs did not appreciably increase (Figure 4). Mucormycosis incidence rates appeared to decrease for the cohorts receiving transplants during the fi rst 4 months (January-April) of each year in the 2003-2005 surveillance period.

Discussion
We describe the modern epidemiology of common non-Aspergillus invasive mold infections at multiple tertiary care sites throughout the United States. During the TRANSNET surveillance period, the estimated cumulative incidence of mucormycosis among HCT subcohorts receiving transplants in 2003 and 2004 was generally higher than in 2002. Public health offi cials and clinicians should be aware of these emerging mold infections, particularly in susceptible hosts, such as transplant recipients.
The worrisome possibility of an increase in mucormycosis incidence has gained attention recently, particularly because this disease has an extremely high case-fatality rate (9,10). Most supporting information about the emergence of this mold has come from compilations of case series (11,12), single-center studies (13)(14)(15), or registries (16,17). One national study in France used administrative data to demonstrate an increase over a 10year period (18). In contrast, our analysis is based on a comprehensive database from a surveillance program that included a large number of centers with a broad geographic distribution. Furthermore, all cases were independently evaluated by using standardized case defi nitions (EORTC-MSG criteria [8]). Reasons for higher incidence of mucormycosis during 2003 and 2004 are not clear from these data. It is certainly possible that these are natural variations or that changes in practices (e.g., mold-active prophylaxis or immunosuppressive agents) or changes in the at-risk populations could be contributory factors. Although we did not see an increase in the numbers of transplants from allogeneic donors performed at the participating sites during this period, we did not have data on other conditions that increase risk for mucormycosis, such as diabetes mellitus or iron overload conditions. The higher incidence during certain periods did not result from a single institution; nor was it likely to be attributable to surveillance artifact, in which improved detection methods falsely suggest a changing trend because other common non-Aspergillus mold infections did not also increase during this period.
Another issue that has been reported recently is the frequency of breakthrough mucormycosis in patients receiving voriconazole (3,4,(19)(20)(21). In these reports, progressive mucormycosis developed in patients who were receiving voriconazole for prophylaxis or treatment for another disease, or these patients were at higher risk for mucormycosis than were control populations (2). In our study, voriconazole was the most frequently reported antifungal drug used before mucormycosis developed, although it was used in fewer than half of all patients. The frequency of voriconazole use before development of mucormycosis is striking, especially in the context of these case reports; however, the implication of the broader use of this antifungal drug to the emergence of this mold is far from clear. Experimental models have demonstrated increased virulence of Mucorales after exposure to voriconazole (22), but other reports have noted an increase of this mold before introduction of voriconazole (14,23). In a randomized trial that used voriconazole as a prophylactic agent, incidence of mucormycosis was not higher among the intervention group than in the group who received fl uconazole (24), although patients meeting enrollment criteria may not have been at high risk for mucormycosis (25). The increase might not be evidence of a causal relationship between voriconazole and mucormycosis but might instead refl ect a parallel increase related to a changing patient risk profi le, including patients at higher risk or increased time at risk, mainly through improved posttransplant survival (25). We did fi nd that 1 factor associated with death in mucormycosis patients on bivariate analysis was prior use of voriconazole (data not shown); this fi nding could indicate that persons receiving voriconazole are more complicated transplant patients and therefore at higher underlying baseline risk for mucormycosis. Because our study did not collect medication data on uninfected controls or on global antimicrobial drug use practices, we were not able to correlate broader voriconazole use to the higher incidence of mucormycosis in this cohort.
We found a high degree of site-to-site variability in the number of cases reported. One hospital contributed 31  nearly double that of the hospital with the second-highest number. No cases were detected at 3 sites. However, the site with 31 cases was also 1 of the sites with the highest number of HCTs performed per year. Whether this variability is a function of environmental or climate conditions, the underlying transplant population at that site, medical practices (e.g., prophylaxis, diagnostic interventions), or a combination of these factors is not known. The variability in the incidence of infections from Mucorales in HCT recipients over time was also intriguing, with the lowest incidence in the cohorts who received transplants during the fi rst 4 months of the year. One recent report described a similar fi nding for aspergillosis among HCT recipients (26). Whether this pattern refl ects an underlying seasonal risk for mucormycosis is not known. However, a seasonal pattern is feasible; these molds exist in the environment, and environmental and temperature conditions can greatly infl uence their growth or sporulation (27). Further study of the seasonal risk for rare mold infections such as these most likely necessitates a larger cohort than that from TRANSNET; large national databases, including those using administrative codes, despite their poor predictive values (28), might be promising for estimating trends for such diseases.
Our data corroborate other reports regarding the clinical features of these mold infections (11,(29)(30)(31)(32)(33). We found that even though most cases occurred soon after transplant, a large number occurred >6 months after transplant. A similar pattern occurs in other mold infections, such as those caused by Aspergillus spp., indicating the need for clinicians to be vigilant about mold infections, particularly a substantial length of time after transplant. Furthermore, we found that the clinical features of these 3 common non-Aspergillus mold infections were similar (with only bloodstream involvement being associated more often with fusariosis, as would be expected) (30). Not surprisingly, the pulmonary system and sinuses were the most common sites of infection for all molds studied, similar to infections caused by Aspergillus spp. These data highlight the similarities in features among mold infections and illustrate the diffi culties in differentiating these infections on the basis of clinical features or organ involvement. Because correct treatment depends on the proper isolate identifi cation, these results reinforce the need for major improvements in diagnostic testing, including development of tissue-based molecular identifi cation methods (34,35) to correctly and effi ciently identify the causative organism.
Our study had a few limitations. Because we collected data only on proven and probable IFIs (and omitted possible infections), as defi ned by EORTC-MSG criteria (8), we are likely to have underestimated disease in these populations. In addition, we were not able to systematically determine which, if any, of these infections resulted from donor-derived infection of the allograft. However, routine posttransplant surveillance conducted by transplant clinical teams, in conjunction with the United Network for Organ Sharing, did not indicate any such clusters.
Our large prospective parallel assessment of the modern epidemiology of these emerging, yet uncommon, non-Aspergillus molds demonstrated a generally higher incidence in mucormycosis in 2003 and 2004 than in 2002 in HCT recipients. More comprehensive and longterm surveillance for these emerging rare molds might be warranted to track trends and assess the changing landscape of these infections.