A single-source nosocomial outbreak of Aspergillus flavus uncovered by genotyping

ABSTRACT During construction work (2017–2019), an increase in Aspergillus flavus infections was noted among pediatric patients, the majority of whom were receiving amphotericin B prophylaxis. Microsatellite genotyping was used to characterize the outbreak. A total of 153 A. flavus isolates of clinical and environmental origin were included. Clinical isolates included 140 from 119 patients. Eight patients were outbreak-related patients, whereas 111 were outbreak-unrelated patients from Danish hospitals (1994–2023). We further included four control strains. Nine A. flavus isolates were from subsequent air sampling in the outbreak ward (2022–2023). Typing followed Rudramurthy et al.(S. M. Rudramurthy, H. A. de Valk, A. Chakrabarti, J. Meis, and C. H. W. Klaassen, PLoS One 6:e16086, 2011, https://doi.org/10.1371/journal.pone.0016086). Minimum spanning tree (MST) and discriminant analysis of principal components (DAPC) were used for cluster analysis. DAPC analysis placed all 153 isolates in five clusters. Microsatellite marker pattern was clearly distinct for one cluster compared to the others. The same cluster was observed in an MST. This cluster included all outbreak isolates, air-sample isolates, and additional patient isolates from the outbreak hospital, previously undisclosed as outbreak related. The highest air prevalence of A. flavus was found in two technical risers of the outbreak ward, which were then sealed. Follow-up air samples were negative for A. flavus. Microsatellite typing defined the outbreak as nosocomial and facilitated the identification of an in-hospital source. Six months of follow-up air sampling was without A. flavus. Outbreak-related/non-related isolates were easily distinguished with DAPC and MST, as the outbreak clone’s distinct marker pattern was delineated in both statistical analyses. Thus, it could be a variant of A. flavus, with a niche ability to thrive in the outbreak-hospital environment. IMPORTANCE Aspergillus flavus can cause severe infections and hospital outbreaks in immunocompromised individuals. Although lack of isogeneity does not preclude an outbreak, our study underlines the value of microsatellite genotyping in the setting of potential A. flavus outbreaks. Microsatellite genotyping documented an isogenic hospital outbreak with an internal source. This provided the “smoking gun” that prompted the rapid allocation of resources for thorough environmental sampling, the results of which guided immediate and relevant cleaning and source control measures. Consequently, we advise that vulnerable patients should be protected from exposure and that genotyping be included early in potential A. flavus outbreak investigations. Inspection and sampling are recommended at any site where airborne spores might disperse from. This includes rarely accessed areas where air communication to the hospital ward cannot be disregarded.

A spergillus flavus is a saprotrophic mold often found in decaying material where it produces readily dispersible conidia (1).It can cause invasive aspergillosis (IA) in neutropenic patients, foremost in regions with arid climates (2,3).It is the second most important cause of IA after Aspergillus fumigatus (4).In general, construction work in and around hospitals, along with contaminated air-supply systems, is the most important risk factor for spreading conidia to patient populations at risk of IA outbreaks (5)(6)(7)(8).However, inhabitation and propagation of A. flavus inside hospitals have also been reported (9).
In 2017-2019, the incidence of IA due to A. flavus increased in children treated for high-risk leukemia at a hematologic cancer center in Copenhagen, Denmark, as previously described (10).The outbreak period coincided with window replacements of the entire ward, which may have caused increased mold exposure.The patients had received amphotericin B (AMB) prophylaxis as per protocol at that time (11), and it was speculated that this regimen selected for A. flavus breakthrough infections (10,12).Outbreak isolates were indeed less susceptible to AMB than A. fumigatus, as expec ted (10,12).All patients recovered from IA with triazole therapy, supplemented with caspofungin until the target plasma concentration was reached.The primary prophylaxis regimen was changed to posaconazole (13) in 2019 with daily caspofungin during antileukemic vincristine treatment.Cleaning and renovation of the ward were intensified.During 2020-2022, the incidence markedly declined, but three additional cases were still observed in patients not on standard protocols.
Aspergillus outbreaks often involve a plethora of different genotypes as air con tains multiple unrelated clones.Nosocomial outbreak strains are rarely isogenic when recovered from patient and environmental samples (7,14,15).However, isogenic nosocomial outbreaks have been described with direct links to in-ward contamination with A. flavus (9,(16)(17)(18)(19).Apparent sources have been linked to water damage in plasterboards and other building materials, dust accumulation on acoustical ceiling tiles, defect air-ventilation filters, etc.
To identify the source of this outbreak, we used microsatellite genotyping of patient and air sample isolates.Microsatellite genotyping is an effective tool for investigating the molecular epidemiology of Aspergillus isolates (14,17,18,20).We included outbreak isolates and comparative isolates from the rest of the country.We found a single-linage A. flavus as the causative agent of this outbreak.A minimum spanning tree (MST) and discriminant analysis of principal components (DAPC) revealed a potentially distinct genetic background of the outbreak clone.All technical risers in the ward were sealed as a consequence of retrieving several isolates from these locations.Repairs of any damaged ward interior, fungal disinfection, and thorough cleaning were done.No A. flavus was found in follow-up air samples.

Setting
The outbreak hospital is a tertiary referral center with approx.1,250 beds and 18,000 discharges a year.The outbreak ward is a general pediatric hematologic-oncologic ward, comprising three medical units with a total of 20 beds (Fig. 1), with rooms for medical staff, ventilation, technical risers, etc. separating crosswise (patient rooms face the windows).The ward is not equipped with high-efficiency particulate air filtration.Hematopoietic stem cell transplantations are performed in a different ward.

Clinical isolates
A total of 140 clinical A. flavus isolates were included, including 14 outbreak isolates from eight outbreak ward patients (2018-2022).A total of 126 were clinical comparator isolates from 111 patients admitted to 14 hospitals or referred via general practitioners .Primary patient specimens (PPS) and pure cultures, previously referred for routine diagnostics at the mycology reference laboratory at Statens Serum Institut, were included.

Control strains
Four external A. flavus reference strains were included: three from an external qualityassessment provider, UK-NEQAS (strain 4903, 5246, and 6719, London, UK) and one from Westerdijk Institute (CBS 128202, Utrecht, The Netherlands).

Environmental sampling
Eighty-six hospital air samples (2 m 3 per sampling site) were collected with a MAS-100NT air sampler (Merck, Søborg, Denmark), from December 2022 to June 2023.Samples were cultured on YGC-agar at 37°C for 2 days to favor the growth of Aspergillus, followed by 3 days at room temperature to facilitate the growth of other fungi for total spore count estimates.Air quality was evaluated using the following categories: areas with critically immunocompromised patients (<1 CFU/m 3 ) (21), areas of patients vulnerable to IA (1 ≤ 5 CFU/m 3 -Prof.Malcolm Richardson, personal communication), no particular risk of IA (5 ≤ 25 CFU/m 3 ), and unacceptable in general (>25 CFU/m 3 ) (21).Identification was based on morphology and MALDI-TOF (Bruker, Roskilde, Denmark) using the MSI spectrum database for molds (22, 23) and when needed (N = 20) supplemented with the sequencing of ITS, BTUB, and CMD (24).
SimpliAmp thermal cycler (TFS) was used for amplification.The program included denaturation (95°C for 10 min), 30 cycles (95°C for 30 s), annealing (58°C for 30 s), elongation (72°C for 30 s), and incubation (72°C for 10 min).PCR of PPSs required 50 reaction cycles.PCR product (1 µL) was added to 13 µL 6.6% GS500-ROX (TFS) size standard/water solution.A denaturation step (95°C for 3 min) followed by instant cooling to prevent reannealing.SeqStudio Genetic Analyzer (TFS) was used for fragment analysis.Peak-Scanner software enabled analyses of fragment peaks (TFS).For conversion of fragment length (bp) to repeat size (STR), we used genome reference data of A. flavus strain CBS 128202 (25).Genetic relations among isolates were visualized with an MST (BIONUMERICS).To detect discriminating patterns among the microsatellite markers (K), a DAPC analysis was run (26) using the R package adegenet 2.1.15(updated 2020).K was selected based on the Bayesian information criteria (BIC-value) and visualized with a scatter plot.Simpson's index of diversity (D-value) was used to estimate the probability of collecting identical genotypes from unrelated sources, including all epidemiologically non-related isolates (neither from the same patient nor as part of the outbreak).

Data analysis
Wilcoxon-test was used for direct comparison of CFU/m 3 for re-tested (paired) locations.Fisher's exact test was used to compare the CFU/m 3 of A. flavus found over time.

Microsatellite typing
Typing of the 140 clinical isolates and four control strains retrieved 116 distinct genotypes (Fig. 2).We calculated a 0.13% probability of detecting identical genotypes among non-related isolates (D-value).One cluster included 23 isolates from 16 different patients, all from the outbreak hospital, and was named "Cluster-1" (Fig. 2 and 3).Genotype and general information on the cluster are shown in Table 1.
All Cluster-1 isolates were obtained during 2018-2022 except one from 2008.Eight were derived from outbreak ward patients, six from cystic fibrosis (CF) patients, and for two, information was not available.We typed isolates of 47 additional patients , from other wards of the outbreak hospital, none of whom harbored the Cluster-1 genotype.

Assessment of indoor air quality
A long-term reservoir within the hospital was suspected on the basis of isogeneity among isolates from the outbreak and the presence of the outbreak clone in the hospital since 2008.As a consequence, air sampling in the outbreak ward was undertaken from December 2022 to January 2023.A total of 258 fungal isolates were cultured from 54 air samples (4.8 CFU/m 3 , range: 0-32.5 CFU/m 3 per sample).The overall fungal CFU/m 3 varied 7.0, 5.1, and 3.3, respectively, among the three dept.(I, III and IV) of the ward.A higher Aspergillus/total-CFU proportion (39.5%, 17/43 CFU) was found in the technical risers compared to the ward (16.67%, 43/258 CFU).The 43 Aspergillus spp.included A. fumigatus (17 isolates), A. flavus (nine isolates), Aspergillus calidoustus (nine isolates), Aspergillus niger (six isolates), and Aspergillus sydowii (two isolates).Of the nine A. flavus isolates found in the air, all had the Cluster-1 genotype.Five of these were found in two of the three technical risers (Fig. 1).

Genetic diversity and further investigations on the outbreak clone
Genetic diversity of the 153 A. flavus isolates was visualized in an MST that shows Cluster-1 as a solitary group in distinction to all other clones (Fig. 2).DAPC analysis identified the main discriminative marker patterns and divided the data into clusters, representing these patterns.The Cluster-1 marker pattern clearly discriminated from all others (Fig. 3)

Infection control and re-assessment of air quality post-intervention
Repairs of any damaged ward interior, fungal disinfection, cleaning of the outbreak ward, and sealing of the technical risers were effectuated by the infection control team.Thirty-two new air samples were collected in areas of dept.I, III, and IV where A. flavus or high CFU/m 3 was found.A total of 104 fungal isolates (3.3 CFU/m 3 ) were isolated, none of which were A. flavus (P = 0.026 comparing 9 A. flavus among 54 samples with 0 A. flavus among 32 samples in the two periods, respectively).The overall CFU/m 3 of fungi for dept.I, III, and IV decreased numerically, from 7.0, 5.1, and 3.3 CFU/m 3 to 2.5, 3.5, and 3.3 CFU/m 3 , respectively (P > 0.5 comparing findings in locations sampled both pre and post intervention).Aspergillus spp.included: A. fumigatus (14 isolates) and A. niger (three isolates).

DISCUSSION
Aspergillus is ubiquitous and thrives in decaying matter.Consequently, construction/ren ovation work with associated release of fungal spores and contaminated air supply systems remain the most frequent causes of Aspergillus outbreaks in hospitals (7).Construction-related outbreaks are most often multi-clonal (14,15,18,27).Thus, we expected limited matches between environmental and clinical isolates in this case as well.However, typing of A. flavus from this outbreak revealed a single-lineage outbreak clone in both clinical and environmental air samples.A possible contamination source could therefore be a specific source of decaying building material, possibly exposed to the air in the process of the window replacement.
Others have found decaying building material to be the source of in-hospital A. flavus propagation.Examples were moistened insulation material of ductwork, plasterboard, acoustical ceiling tiles, and also dust (9,16).This was the case in a study from reference 16, which reported an outbreak of IA in a cardiac surgery ward caused by A. flavus (16,28).Air and surface samples were collected, and genetic isogeneity was found in all isolates from environmental and clinical samples.Subsequent to that outbreak, air/surface sampling disclosed an increasing gradient of A. flavus isolates toward the focal point of contamination, apparently water damage beneath the ward floor (16).Other examples exist where matching genotypes between clinical and environmental A. flavus isolates have been found (15,17,19).One study used a similar microsatellite panel to ours and found a cluster of four patients and four environmental isolates from a bronchoscopy ward, where all four patients had been treated (17).These studies support the legitimacy of our findings of a single predominant outbreak clone.
In our study, the earliest outbreak clone isolate was found in a patient in 2008, which indicates the presence of the lineage for more than a decade.Conidia from the outbreak clone may have lingered in the outbreak ward for many years without access to patients.Conidia contained in dust may have contaminated the outbreak ward during the window replacement, thus leading to the outbreak.Conidia possibly settled in areas with no regular cleaning routine, like the technical risers.This is supported by the repeated finding of the outbreak clone in the technical risers and the inability to find it in the post-intervention air samples after the technical risers had been sealed.Technical risers could potentially also contain particularly suitable material for fungal growth, as the risers house hot-water piping offering suitable propagation temperature.Sealing and cleaning of the risers were requested as part of the infection control measurements, in the aftermath of our findings.Unfortunately, this prohibited further sampling of the riser's interior.
Several studies have affirmed microsatellite typing as an effective tool for discrimi nation between different lineages of molds within a population (4,14).Stability, low costs, and high discriminative power among non-related isolates are large benefits.Indeed our calculation of D-value rendered a similarly high discriminative value (0.9987), which corresponds well with other studies using a similar microsatellite panel (17,20).Furthermore, a high interlaboratory reproducibility gives this method a clear advantage compared to other typing methods (4).The MST (Fig. 2) visually reflected the linkage of Cluster-1 isolates within the A. flavus population.The fact that this lineage was found solely in patients from the outbreak hospital and multiple times in the hospital environment gives epidemiological evidence of a single-source nosocomial outbreak.However, DAPC analysis (Fig. 3) revealed the genetic distinction of the outbreak clone genotype compared to all other A. flavus isolates.This suggests a different genetic inheritance of the outbreak clone compared to other A. flavus clones.This raised the question of whether the outbreak clone had a competitive genetic trait.Of note, no other genotypes of A. flavus were found in the hospital air, and this isogeneity does not merely reflect rarity in culturing A. flavus from air samples in outbreak settings (9,15,29).It may indicate an advantage in adaptability to survive in the hospital environ ment compared to other A. flavus clones.WGS-based typing would be interesting to further elucidate this relationship and to discover if this outbreak clone might even be a subspecies under the A. flavus complex.
In conclusion, we learned that one specific lineage of A. flavus clones lingers in the outbreak ward, and it has a distinct microsatellite marker pattern compared to other A. flavus clones.We speculate whether this has any influence regarding virulence, pathogenicity, or adaptability.However, the outbreak initially described by reference (10) showed no mortality from the outbreak itself.An unsuspected outcome compared to other outbreaks, with high mortality in similar patient populations (7).This might point toward a lower virulence.Air sampling was useful inside the outbreak ward, and we managed to detect the outbreak clone inside the ward itself.Besides genotyping during outbreaks of IA, we emphasize the need for regular cleaning that includes obscure sites like technical risers where contamination of patient environments cannot be excluded due to maintenance.We aim to further monitor the hospital environment and investi gate this particular outbreak clone to better understand its nature.

FIG 1
FIG 1 Floorplan of the outbreak ward. A. flavus was found in dept.I, III, and IV.The number inside the circles indicates A. flavus colony forming units (CFU) found in air samples, in each location.Technical risers (red) are vertical corridors for water piping, running from the basement to the building roof.Unsealed cabinet doors separate them from the wards.

FIG 2
FIG 2 MST of 144 clinical A. flavus isolates and nine air-sample isolates.Each node is a multi-locus genotype (MLG), with variable size depending on the number of isolates within.Node distance represents genetic distinction among MLGs.Connecting lines denote a number of varying markers, thick black is a one-marker variation, thin solid is a two-marker variation, dashed is a three-marker variation, etc. Light/dark blue nodes are outbreak hospital (OH) MLGs.Cluster-1 is highlighted.

FIG 3
FIG 3 DAPCs, in simple terms, a mathematical model representing the most discriminative differences in the genetic marker patterns, of all isolates in our population.Based on the BIC-value, five clusters credibly show the significant outlying of Cluster-1.Eigenvalues of D1/D2 assure the use of a 2D plot to display almost all discrimination between the clusters (left-hand corner).Clusters 2-5 discriminate mainly along D2, and Cluster-1 discriminate along D1.

TABLE 1
Patients in Cluster-1 and the outbreak clone genotype a First described in 2021 by Vissing et al. as part of an IA outbreak.(10).b Cystic fibrosis patients.c Outbreak ward patients.d Bold font marks deviation from overall genotype.