A total of 146 Scedosporium isolates from 120 episodes (119 patients) were studied (Technical Appendix). Demographic data were available for 108 (90%) episodes and coexisting conditions and risk factor data for 115 (95.8%). Most episodes were reported from New South Wales (64.2%), followed by Victoria (19.2%) and Western Australia (9.2%). The male: female ratio was 1.3: 1. The major patient coexisting conditions and known risk factors for scedosporiosis are summarized in the Technical Appendix. Thirty-nine patients (32.7%) had no underlying medical condition. Coincident building construction was noted in 27 cases (22.5%). Scedosporium isolates were associated with invasive disease in 46 (38.3%) instances; the remaining 74 (61.7%) were isolated from patients who were colonized (Table 2).

We examined genetic variation among a large number of population-derived Scedosporium isolates across the Australian continent. In line with previously reported genetic variability in the S. apiospermum/P. boydii species complex (30–32), we observed 2 distinct ITS-RFLP patterns among S. apiospermum isolates, showing the presence of the newly described species S. aurantiacum (11). Notably, we have identified by ITS sequencing that S. aurantiacum comprised 45% of the current collection of Australian “S. apiospermum” isolates and documents genetic variability within S. aurantiacum.

Epidemiologic investigation of Scedosporium infection requires accurate identification and typing. S. apiospermum, S. aurantiacum, and S. prolificans were clearly distinguished from each other by PCR fingerprinting and ITS-RFLP analysis. This is consistent with previous rDNA sequence-based studies (30,33,34). The observation of 2 distinct genetic groups, corresponding to S. aurantiacum and S. apiospermum, supports the proposal that S. aurantiacum be designated a separate species (11). This proposal is also supported by the 5%–10% ITS sequence variation found between S. aurantiacum and S. apiospermum compared to an absence of intraspecies variation in S. aurantiacum and S. prolificans and a 2.2% variation in S. apiospermum ( 30,32; current study).

Using PCR fingerprinting, intraspecies variation was greatest (58%) among S. apiospermum isolates (Figure 3). This diversity is generally consistent with the high degree of polymorphism (15–20 genotypes) previously found (10,20,32). In contrast, genetic variation was lowest (28%) among the S. aurantiacum isolates (Appendix Figure 1; Figure 3). Nevertheless, PCR fingerprinting polymorphisms clearly differentiated all 30 strains (Appendix Figure 1). Further genotyping studies of a greater number of and more geographically diverse S. aurantiacum isolates are warranted.

The intraspecies PCR fingerprint variation in S. prolificans (45%) was greater than that in S. aurantiacum but less than that in S. apiospermum. Given that S. aurantiacum is phylogenetically more closely related to S. apiospermum than to S. prolificans ( 11,33; current study), this result was unexpected. It may be due to different evolutionary pressures acting on the 3 different species or the relatively small numbers of S. aurantiacum isolates studied to date. The moderate genetic diversity among S. prolificans confirms previous findings (19). Despite the observed polymorphisms, PcoA of PCR fingerprint profiles showed dense clustering for S. prolificans (Figure 3), which is consistent with the low to absent intraspecies variability in S. prolificans found by others (20,21,33). These apparently contradictory findings emphasize the importance of choosing the optimum molecular typing tool with the most appropriate discriminatory power for the organism or species being studied.

The high degree of intraspecies variation detected by PCR fingerprinting and AFLP analysis supports the use of these methods to establish genetic relatedness between isolates recovered from different patients or multiple isolates from the same patient. In comparison, the variation detected by ITS-RFLP analysis and ITS sequencing corresponded to interspecies variation, which makes those techniques ideal for identification of any given isolate to the species level. Individual patients are most likely infected or colonized with genetically distinct strains ( 19–21; this study). Identical PCR fingerprint or AFLP profiles were noted in multiple isolates recovered simultaneously from different anatomic sites in the same patient (21; current study). However, 8 patients were infected or colonized by at least 2 strains as reflected by their different genetic profiles (Technical Appendix). Possible explanations include concomitant infection by multiple strains from which only a restricted number were recovered, or colonization by 1 strain followed by infection or colonization with a second strain of a different genotype. Longitudinal genotyping studies are required to determine the likelihood that persistence of >1 genotypes later leads to clinically important infection or whether the disease is more likely to be caused by an unrelated genotype. In this context, the development of a multilocus sequence typing scheme for Scedosporium, as has been developed for Candida spp. (35), would be of great advantage to overcome interlaboratory reproducibility problems, which are known to be associated with PCR fingerprinting or AFLP data. However, developing such a scheme remains cumbersome due to the current lack of genomic data of Scedosporium spp.

For all 3 Scedosporium spp., there was no clustering of strains according to their geographic or body site of origin or by their ability to cause invasive disease, which is in agreement with previous findings for S. apiospermum (20,30) and S. prolificans (16,17,21). Of note, no specific genotypes were associated with underlying medical conditions or risk factors. Compared with S. apiospermum and S. aurantiacum S. prolificans was more frequently associated with coincident hospital renovation, and invasive disease, had a greater predilection to cause disseminated infection and was the predominant species isolated from blood and other sterile sites (12–16,36; current study). Our preliminary observations indicate that the epidemiology and clinical relevance of recovering S. aurantiacum may be similar to that of S. apiospermum. S. aurantiacum has been reported to colonize the respiratory tract of at-risk patients (8).

In addition to PCR fingerprinting, we applied AFLP analysis to investigate the possibility of 2 case clusters caused by S. prolificans. AFLP analysis was chosen as an independent technique using 2 combinations of selective primers (Table 1), which have been previously shown to have good discriminatory power for fungal strain differentiation (26). Both techniques, previously used to identify outbreak strain clusters in the recent cryptococcosis outbreak on Vancouver Island (37), generated in the current situation distinct patterns from all S. prolificans isolates except serial isolates obtained from the same patient (Appendix Figure 1, Appendix Figure 2). These findings exclude the occurrence of nosocomial outbreaks or any close relationship with the nonoutbreak isolates, a result similar to those obtained previously (38). Overall nosocomial acquisition of infection has been demonstrated in only 2 instances (16,17). Scedosporium spp. have rarely been isolated from hospital air or from indoor or outdoor surface samples (13,39,40, current study), which raises questions about the mode of acquisition by patients and the mechanisms of the selection of this specific fungus as an infectious agent from among the high biodiversity of environmental molds.

In conclusion, ITS-RFLP analysis is a powerful tool for distinguishing between isolates of the new species S. aurantiacum and S. apiospermum. PCR fingerprinting and AFLP analysis are useful techniques for determining genetic relatedness between Scedosporium isolates and for investigating potential case clusters.