Assessment of the molecular identification algorithm and its impact on antifungal susceptibilities against clinical Fusarium isolates: a multicentre study in Taiwan, 2011–2023

Abstract

Objectives

Invasive fusariosis is rising and poses challenges due to species complexity and antifungal resistance. In vitro susceptibilities of new antifungals against Fusarium isolates are seldomly evaluated in Asia. This study aimed to evaluate a two-step molecular identification algorithm and to perform in vitro antifungal susceptibility with correlation of species and susceptibility patterns.

Methods

Fusarium clinical isolates collected at three hospitals in Taiwan (2011–2023) were identified to species level using sequential ITS and TEF1α sequencing (step I), followed by RBP2 sequencing (step II) for inconclusive isolates. Minimum effective/inhibitory concentrations (MECs/MICs) of manogepix, olorofim, amphotericin B and voriconazole were determined by EUCAST method (E.Def 9.4).

Results

Of 103 isolates (37 blood and 66 cornea isolates) evaluated, the two-step algorithm achieved >90% to species level. Fusarium solani species complex (FSSC) was predominant, especially in blood isolates (86.5% versus 65.2% in cornea isolates; P = 0.02). The rest 28 isolates belonged to 12 species within six species complexes (SCs). Manogepix exhibited potent activity against all isolates (MEC ≤0.015 mg/L), while olorofim activities varied by SCs, with MIC ≤0.25 mg/L against Fusarium fujikuroi SC. FSSC displayed higher voriconazole and amphotericin B MICs compared with other SCs, with Neocosmospora keratoplastica displaying a highest amphotericin B modal MIC of 4 mg/L. Four major Neocosmospora species showed voriconazole MIC ≥16 mg/L.

Conclusions

Our findings indicated the two-step molecular algorithm accurately identifies Fusarium to species level. Further, we underscored the significance of considering both Fusarium SCs and species for predicting antifungal susceptibility, particularly to olorofim and amphotericin B.

Introductions

Fusarium is listed as one of the WHO high priority fungal pathogens,¹ reflecting its emergence within One Health frameworks and the current lack of effective antifungal agents. This genus is globally distributed across diverse environments and human habitats, serving as reservoirs to threaten health of human, animals and plants. Fusarium can be not only pathogens damaging crop widely and causing up to 50% yield loss in soybean, banana, tomato and wheat,² but also cause human infections in the community and healthcare settings due to environment exposure. Most pathogenic species have been detected in environmental samples.³ Over 20 species complexes (SCs) exist in nature, with at least 7 SCs linked to human infections, ranging from superficial (such as keratitis and onychomycosis), locally invasive (cellulitis, sinusitis and pneumonia), to disseminated infections with positive blood cultures.^4,5

Conventional antifungal agents offer limited efficacy against Fusarium; notably, Fusarium solani SC (FSSC) exhibit intrinsic resistance to azoles.^6,7 Among different Fusarium SCs, susceptibilities to polyenes are variable. Among novel antifungal agents, manogepix targets Gwt1, a critical enzyme for synthesizing the glycosylphosphatidylinositol-anchored proteins of the fungal cell wall, and demonstrates notable in vitro activity against various Fusarium SCs.^8–10 Olorofim, which inhibits dihydroorotate dehydrogenase, essential for the de novo pyrimidine synthesis pathway, may also exhibit in vitro activity against selected SCs.^11,12 Yet, comparative data on the antifungal susceptibility profiles among species within the same SC remain scarce. In addition, routine implementation of antifungal susceptibility testing for Fusarium in clinical laboratories is challenged by technical limitations. There is ongoing interest in determining whether precise species identification can reliably predict in vitro susceptibility to novel and conventional antifungal agents.

Accurate species identification within a given Fusarium SC remains challenging due to morphological similarities that prevent accurate differentiation by conventional techniques. Furthermore, incomplete MALDI-ToF databases hinder identification capabilities in clinical laboratories.^13,14 To overcome these limitations, molecular identification offers a robust alternative for species identification. While ITS is useful in the discrimination between the multiple SC of Fusarium, it lacks discriminatory power down to species level. Consequently, TEF1α is the preferred first-line marker for species identification. Also, it has higher PCR amplification and success rates compared with RPB2. Therefore, the current European Confederation of Medical Mycology and International Society for Human and Animal Mycology (ECMM/ISHAM) guideline recommends molecular identification using ITS and TEF1α for Fusarium.¹⁵ Although RPB2 sequencing can resolve specific ambiguities of TEF1α—particularly within the FSSC and F. fujikuroi species complex (FFSC)—it remains uncertain whether additional taxonomic resolution correlates with a species-specific resistance pattern. We propose an efficient two-step molecular algorithm: initially sequence ITS and TEF1α (Step I), and, if species-level identification is inconclusive, sequence RPB2 (Step II).

To address these gaps, we conducted a multicentre study to validate the two-step molecular algorithm for Fusarium species identification. We also compared distributions of Fusarium species identified by molecular methods from isolates causing fungaemia and cornea infections. In addition, we analysed in vitro susceptibilities of two antifungal agents under phase III clinical trials (manogepix, olorofim) and two currently preferred agents (amphotericin B, voriconazole) at both SC and individual species levels, aiming to determine the correlations between antifungal susceptibilities and Fusarium species.

Methods

Isolates

Fusarium clinical isolates were prospectively collected from patients at three hospitals between 2011 and 2023. The isolates were obtained from blood and cornea specimens, with two exceptions: one from an anterior chamber aspirate and one from contaminated contact lens solution. The following mycological studies were conducted at a reference laboratory at National Taiwan University Hospital (NTUH) after confirmation as Fusarium species and related genera by macroscopic colony morphology and microscopic features in a lactophenol wet mount preparation according to standard laboratory procedures. To account for potential epidemiological variations driven by geoclimatic differences, we stratified the study population into two geographically distinct groups: isolates from NTUH and NTUH Cancer centre located in subtropical Taiwan as a discovery cohort, and used isolates from National Cheng Kung University Hospital located in tropical Taiwan as a validation cohort.

Sequencing for molecular identifications

Fusarium isolates were cultured on Sabouraud dextrose agar (Liofilchem, Via Scozia, Italy). Culture plates were incubated at 28°C and grown for up to 7 days. DNA extraction was performed by Quick-DNATM Fungal/Bacterial Miniprep Kit (Zymo Research, Irvine, USA). Three gene regions, ITS, TEF1α and RPB2, were amplified directly from the genomic DNA for sequencing using primer pairs ITS5/ITS4,¹⁶ EF1/EF2¹⁷ and RPB2–5f2, fRPB2–7cr, fRPB2–7cf and RPB2–11ar.^18,19 The PCR conditions and primers were described and listed in the Supplementary Materials (available as Supplementary data at JAC-AMR Online) and Table S1, respectively.

All sequence results were uploaded and aligned to reference sequences deposited in the public databank (https://www.fusarium.org/) to confirm the species level of isolates tested. Based on the two-step algorithm, step I to achieve species identification was by using the nucleotide sequences of ITS and TEF1α. If both of ITS and TEF1α cannot determine a single species for a specific isolate, step II would use the full nucleotide sequence length of RPB2 for species identification. If the alignment results in step II were unable to achieve single-species identification, the reporting rules were as follows: both species would be reported if only two species within the same species complex identified; otherwise, a species complex would be reported if ≥3 species within the species complex identified.

Antifungal susceptibility testing

Broth microdilution methods for minimum effective concentrations (MEC) of manogepix (Cat no. HY-18233, MedChemExpress), and minimum inhibitory concentrations (MIC) of olorofim (Cat no. HY-104029, MedChemExpress), amphotericin B (Sigma-Aldrich) and voriconazole (Sigma-Aldrich) against all isolates were performed according to EUCAST E.Def 9.4 susceptibility testing with a final inoculum size of 1–2.5 × 10⁵ cfu/mL,^9,20 except filtration (11-µm filter) of the inoculum was performed only if a significant number of hyphae are detected (>5% of fungal structures). The results were interpreted by EUCAST: tentative epidemiological cut-off values of amphotericin B for FFSC and FSSC were both ≤8 mg/L. A. flavus ATCC 20430404304 and A. fumigatus ATCC 20430504305 were used as controls.

Statistics

MIC/MEC ranges, modal MIC and MIC₅₀ and MIC₉₀ (the MIC value that includes 50% and 90% of the isolates, respectively) values were calculated. Categorical variables were expressed by numbers (percentages) and compared by using the chi-square test with Bonferroni-adjusted α for pair-wise comparisons as post hoc analysis. A two-sided P value <0.05 was considered significant. All statistical analyses were performed using Stata software (version 17; StataCorp, College Station, TX, USA).

Results

Performance of the two-step molecular algorithm for Fusarium

Of 103 isolates collected from 96 patients evaluated, 37 (35.9%) were obtained from blood samples and 66 (64.1%) were isolated from cornea specimens. Notably, the proportions of blood isolates increased substantially over 13 years, from 0% to 64.3% (P for trend <0.001, Figure 1). In the discovery cohort (n = 45), initial step I identification classified 38 isolates (84.4%) to species levels, with step II increasing species-level identification by an additional 13.3% (n = 6, Figure 2a). Comparable results were observed in the validation cohort, with step I identifying most species level and an ≈9% increase after step II. After the two-step identification, 2.2% and 6.9% of isolates remained unidentified in the discovery and validation cohorts, respectively. The proportions of identification in each step were consistent between blood and cornea isolates (Figure 2b).

Figure 1.

Distributions of Fusarium clinical isolates stratified by study cohort and specimens, and the trend of proportions of blood isolates during 2011 and 2023. D, discovery cohort; V, validation cohort.

Figure 2.

Distributions of Fusarium clinical isolates confirmed to species levels based on a two-step molecular identifications algorithm by study cohort (a), specimens (b) and SCs (c). ITS, internal transcribed spacer region of the nrDNA; RPB2, RNA polymerase second largest subunit; TEF1α, translation elongation factor 1 alpha.

Distributions of species complexes and species in Fusarium

FSSC was the predominant SC, accounting for 32 (78.4%) and 43 (65.2%) isolates in blood and cornea samples, respectively (Table 1). Overall, FFSC (n = 8, 7.8%) and Fusarium dimerum species complex (FDSC, n = 7, 6.8%) shared the second and third common species complex, followed by Fusarium incarnatum–equiseti species complex (FIESC, n = 6), Fusarium oxysporum species complex (FOSC, n = 4) and other SCs (n = 3). After two steps of molecular identification all but one FSSC isolates were identified to species level, while 14.3% of non-FSSC isolates (4/28) remained classified only at SC level (P < 0.001) (Figure 2b). The proportion of FSSC in blood isolates were significantly greater than that of in cornea isolates [86.5% (32/37) versus 65.2% (43/66); P = 0.02].

Table 1.

Distributions of Fusarium species by molecular identifications among blood and cornea specimens

Source	Species complexa	Species	Discovery cohort (n = 45)b	Validation cohort (n = 58)b	Total (n = 103)
Blood			33 (73.3)	4 (6.9)	37 (35.9)
	FSSC (n = 32)	Neocosmospora bataticola	1 (2.2)	0 (0)	1 (1.0)
		Neocosmospora bostrycoides	2 (4.4)	0 (0)	2 (1.9)
		Neocosmospora diminuta	1 (2.2)	0 (0)	1 (1.0)
		Neocosmospora falciformis	1 (2.2)	0 (0)	1 (1.0)
		Neocosmospora ipomoeae	1 (2.2)	0 (0)	1 (1.0)
		Neocosmospora keratoplastica	7 (15.6)	0 (0)	7 (6.8)
		Neocosmospora petroliphila	5 (11.1)	2 (3.4)	7 (6.8)
		Neocosmospora pseudensiformis	11 (24.4)	0 (0)	11 (10.7)
		FSSC (unknown species)	1 (2.2)	0 (0)	1 (1.0)
	FFSC (n = 2)	Fusarium napiforme	1 (2.2)	0 (0)	1 (1.0)
		Fusarium planum	1 (2.2)	0 (0)	1 (1.0)
	FDSC (n = 2)	Bisifusarium dimerum	1 (2.2)	1 (1.7)	2 (1.9)
	FIESC (n = 1)	FIESC (unknown species)c	0 (0)	1 (1.7)	1 (1.0)
Cornea			12 (26.7)	54 (93.1)	66 (64)
	FSSC (n = 43)	Neocosmospora falciformis	5 (11.1)	20 (34.5)	25 (24.3)
		Neocosmospora ferruginea	1 (2.2)	0 (0)	1 (1.0)
		Neocosmospora keratoplastica	3 (6.7)	10 (17.2)	13 (12.6)
		Neocosmospora metavorans	2 (4.4)	2 (3.4)	4 (3.9)
	FFSC (n = 6)	Fusarium annulatum	0 (0)	2 (3.4)	2 (1.9)
		Fusarium erosum	0 (0)	1 (1.7)	1 (1.0)
		Fusarium proliferatum	0 (0)	1 (1.7)	1 (1.0)
		Fusarium pseudocircinatum	0 (0)	2 (3.4)	2 (1.9)
	FIESC (n = 6)	Fusarium mucidum/Fusarium aberrans	0 (0)	1 (1.7)	1 (1.0)
		Fusarium tanahbumbuense	1 (2.2)	1 (1.7)	2 (1.9)
		FIESC (unknown species)d	0 (0)	3 (5.2)	3 (2.9)
	FDSC (n = 5)	Bisifusarium delphinoides	0 (0)	2 (3.4)	2 (1.9)
		Bisifusarium lovelliae	0 (0)	2 (3.4)	2 (1.9)
		Bisifusarium nectrioides	0 (0)	1 (1.7)	1 (1.0)
	FOSC (n = 4)	Fusarium cugenangense	0 (0)	2 (3.4)	2 (1.9)
		Fusarium liriopes	0 (0)	1 (1.7)	1 (1.0)
		Fusarium nirenbergiae	0 (0)	1 (1.7)	1 (1.0)
	FDECSC (n = 1)	Albonectria rigidiuscula	0 (0)	1 (1.7)	1 (1.0)
	FNSC (n = 1)	Fusarium commune	0 (0)	1 (1.7)	1 (1.0)

FDECSC, Fusarium decemcellulare species complex; FNSC, Fusarium nisikadoi species complex.

^a P for SCs distributions between discovery and validation cohort, 0.01; for those between blood and cornea, 0.25.

^bIsolate numbers and percentages in bold indicate the top species of blood and cornea samples in the two cohorts, respectively.

^cCould be Fusarium sulawesiense, F. pernambucanum, F. caatingaense, F. irregulare, F. multiceps, F. luffae, F. annulatum or F. persicinum.

^dNCK0034 could be Fusarium sulawesiense, F. pernambucanum or F. caatingaense. NCK1044 and NCK1070 could be Fusarium sulawesiense, F. pernambucanum, F. caatingaense, F. irregulare, F. multiceps, F. luffae, F. annulatum or F. persicinum.

Within FSSC, blood isolates demonstrated the relatively even distribution among Neocosmospora pseudensiformis (34.3%, 11/32), Neocosmospora keratoplastica and Neocosmospora petroliphila (both of each, 21.9% [7/32]), while cornea isolates showed Neocosmospora falciformis (58.1%, 25/43) as the predominant species, followed by N. keratoplastica (30.2%, 13/43). By contrast, non-FSSC species in cornea samples exhibited extensive diversity: FFSC belonged to four species (n = 6), FDSC belonged to three species (n = 5), FIESC belonged to at least two species (n = 6) and FOSC belonged to three species (n = 4).

In vitro susceptibilities of antifungal agents against Fusarium

The distribution of amphotericin B MICs ranged widely (0.125–>16 mg/L) and varied by SCs with variable modal MICs (0.5–2 mg/L) in four major SCs and highest MIC₉₀ for FIESC (>16 mg/L) (Table 2). In vitro activities of voriconazole were very poor against all SCs, with modal MICs ≥4 mg/L and all MIC₉₀ > 16 mg/L. Manogepix exhibited potent and consistent activity across all 103 isolates, with MEC values at the lowest tested concentrations of ≤0.015 mg/L in each of SCs. Olorofim MICs also ranged widely and varied by SCs. Olorofim was highly active against FFSC with a model MIC of 0.008 mg/L, while its MIC₅₀/MIC₉₀ values for FSSC, FDSC and FIESC were higher than the upper limit of the tested concentrations, 0.5 mg/L.

Table 2.

Antifungal minimum inhibitory/effective concentrations against Fusarium species by SCs

	Amphotericin B MIC (mg/L)			Voriconazole MIC (mg/L)			Manogepix MEC (mg/L)			Olorofim MIC (mg/L)
	Modala	MIC50/MIC90a	Range	Modala	MIC50/MIC90a	Range	Modala	MEC50/MEC90a	Range	Modala	MIC50/MIC90a	Range
FSSC (n = 75)	1	1/4	0.125–>16	>16	>16/>16	0.06–>16	≤0.015	≤0.015	≤0.015	>0.5	>0.5/>0.5	>0.5
FFSC (n = 8)	2	2/4	0.5–4	4	4/>16	1–>16	≤0.015	≤0.015	≤0.015	0.008	0.008/0.25	0.008–0.25
FDSC (n = 7)	0.5	0.5/2	0.25–2	16	4/>16	2–>16	≤0.015	≤0.015	≤0.015	>0.5	>0.5/>0.5	>0.5
FIESC (n = 6)	1	2/>16	1–>16	8	8/>16	0.125–>16	≤0.015	≤0.015	≤0.015	>0.5	>0.5/>0.5	>0.5
FOSC (n = 4)	NAa	NAa	1–8	NAa	NAa	8–>16	NAa	NAa	≤0.015	NAa	NAa	0.25–>0.5
Others (n = 3)8	NAa	NAa	0.5–>16	NAa	NAa	1–>16	NAa	NAa	≤0.015	NAa	NAa	>0.5

MIC₅₀ and MIC₉₀ represent the lowest concentration of the antifungal at which 50% and 90% of the isolates were inhibited, respectively; NA, not applicable.

^aModal MIC/MEC, MIC₅₀/MEC₅₀ and MIC₉₀/MEC₉₀ were calculated for SCs with ≥5 isolates.

^bThe other group included each of the isolates belonging to the Fusarium decemcellulare species complex, F. nisikadoi species complex and Fusarium species.

Overall, FSSC generally displayed higher MICs for antifungal agents tested except manogepix. Of note, the ranges of amphotericin B and voriconazole MICs against FSSC distributed widely (0.125–>16 mg/L, and 0.06–>16 mg/L, respectively). Thus, we further analysed antifungal susceptibilities of these two preferred available antifungal agents to species level within FSSC (n ≥ 5 per species). N. keratoplastica, the second most common species, had the highest modal MIC for amphotericin B at 4 mg/L, while those of N. falciformis, N. pseudensiformis and N. petroliphil were 0.5–1 mg/L (Table 3). Nevertheless, the proportion of non-wild-type isolates to amphotericin B was greatest in N. pseudensiformis (18.2%), compared with N. keratoplastica (5.0%), N. falciformis (3.8%) and N. petroliphila (0%), resulting in similar geometric means (GMs) MICs for N. pseudensiformis and N. keratoplastica (3.31 versus 3.03 mg/L). For voriconazole, most isolates in four common Neocosmospora species showed MICs ≥16 mg/L.

Table 3.

Distributions of antifungal MIC against common Neocosmospora (n  ≥  5) by species

	MICs (mg/L)a
	≤0.03	0.06	0.12	0.25	0.5	1	2	4	8	≥16	GM	Non-WT, n (%)b,c
N. falciformis (n = 26)
Amphotericin B	0	0	0	0	7	12	5	1	0	1	1.14	1 (3.8)
Voriconazole	0	0	0	0	0	0	0	1	0	25	≥16	NA
N. keratoplastica (n = 20)
Amphotericin B	0	0	0	0	0	2	7	9	1	1	3.03	1 (5.0)
Voriconazole	0	0	0	0	0	0	0	0	0	20	≥16	NA
N. pseudensiformis (n = 11)
Amphotericin B	0	0	1	0	0	5	2	1	0	2	3.31	2 (18.2)
Voriconazole	0	0	0	0	0	0	0	0	0	11	≥16	NA
N. petroliphila (n = 8)
Amphotericin B	0	0	0	0	4	1	1	2	0	0	1.22	0 (0)
Voriconazole	0	0	0	0	0	0	0	0	0	8	≥16	NA

GM, geometric means; NA, not applicable.

^aIsolate numbers in bold indicate modal MICs.

^bThe minimum inhibitory concentration results were interpreted by EUCAST: tentative epidemiological cut-off values of amphotericin B for FFSC and FSSC were both ≤8 mg/L.

^c P for overall comparisons of non-WT type proportions, 0.31.

Discussions

This current study is the largest series of Fusarium clinical isolates incorporating those obtained from blood cultures (Table S2) and demonstrates a two-step molecular identification protocol facilitating precise species-level classification in >90% of isolates evaluated. We identified FSSC to be the major pathogenic SC overall, and more commonly found in blood than cornea samples among seven SCs. Besides, the species distributions of FSSC varied by clinical samples. Of four antifungal agents evaluated, manogepix exhibited potent activity against all isolates, while olorofim activities varied by SCs. MICs of amphotericin B and voriconazole varied at SC and species levels. Overall, FSSC generally exhibited higher MICs for amphotericin B and voriconazole compared with non-FSSC. All these findings not only reinforced the robustness of current international guidelines for laboratory identification of this clinically relevant genus, but also elucidates the correlations between antifungal susceptibilities and Fusarium species, particularly for FSSC.

The molecular identification of Fusarium by ITS and the TEF1α gene is currently recommended by international medical guidelines and agriculture societies, while RPB2 allows for enhanced discrimination between closely related species of FFSC and FSSC.^15,21,22 However, previous clinical studies focused on analyses of molecular identification of Fusarium by the TEF1α gene only (Table S2).^{13,14,23–25} This study is among the first, to best of our knowledge, of the clinical studies to test the molecular identification of Fusarium from clinical samples by two steps encompassing ITS, TEF1α and RPB2. The decision to incorporate RPB2 sequencing was driven by its ability to resolve the closely related species within FSSC and FFSC that remain undistinguished by TEF1α alone. Uniquely, this study quantified the clinical yield of this approach: while ITS and TEF1α correctly identified >80% of clinical isolates, the second step of RPB2 sequencing successfully identified a further 10% (11/103). Notably, this added resolution was not limited to FSSC (n = 4), but also identified species within FIESC (n = 3), FDSC (n = 2), FFSC (n = 1) and FOSC (n = 1) (Figure 2c). Our results were consistent in both discovery and validation cohorts, supporting the recommendations by current guidelines and consensus to use RPB2 for resolving ambiguities in TEF1α analysis.^15,21,22

In this study, most clinical Fusarium isolates belonged to FSSC, aligning with previous reports from tropical or subtropical countries, including two studies conducted in Taiwan.^14,23–28 The prevalence of specific Fusarium SCs varied by epidemiological factors. First, geographic variations matter (Table S2). Specifically, FSSC is dominant in Asia and South America, in addition to latitudinal differences.^14,23–29 In Europe and the USA, two different multicentre studies found FOSC is the leading SC (35.7% and 73.8%, respectively).^11,13 Second, the sample sources differ, resulting in variations of the distributions of Fusarium SCs even under the similar climatic condition. For instance, a multicentre study in France reported the most common SC isolated from blood as over a 10-year period was FFSC (53.7%, 29/54), rather than FOSC.³⁰ In addition, our study found the proportions of FSSC differed between blood and cornea samples. Third, patient populations also differ. A multicentre study from Spain between 2000 and 2015 indicated non-neutropenic patients often have localized fusariosis with dominant unknown Fusarium SCs (63.2%), while neutropenic patient had more frequently have disseminated fusariosis with FSSC as the top SC (38.5%).³¹ These observations highlight the need for ongoing monitoring of clinical Fusarium epidemiology, considering patient groups, sample types and regional differences.

On the other hand, ocular trauma represents the most common prevalent cause of Fusarium keratitis in settings without contact lens-related outbreaks, with patients often exposed to Fusarium via contaminated soil or plants in nature.^24,26,27 In this study, a total of 23 isolates obtained from cornea samples were clustered to 14 species within six non-FSSC, highlighting the extensive diversity of Fusarium in the environment. Of note, we identified the first human keratitis caused by Albonectria rigidiuscula at a hospital located in tropical Taiwan. This species, belonging to Fusarium decemcellulare species complex, is recognized as an important phytopathogen affecting common agriculture plants, such as mango, in tropical areas including Taiwan.^21,22 Given Fusarium consists of at least 300 phylogenetically distinct species and 23 SCs,^21,22 our result underscored that Fusarium is a continuous threat to human health from a One Health perspective.

As for in vitro susceptibility testing, manogepix exhibited consistent activity against a range of Fusarium SCs in this study, aligning with previous findings obtained using either the EUCAST or CLSI methodologies.^8–10,32 Although earlier studies were conducted primarily in the USA and Europe, the Asia–Pacific area remains critical for Fusarium infections, as shown in Table S2. This report systemically investigates by manogepix MECs using the EUCAST method, providing comprehensive and timely data that show manogepix MECs at the lowest tested concentrations across all Fusarium isolates. These results support the implementation of the ongoing phase III clinical trial of fosmanogepix, a prodrug of manogepix, for treatment of adult patient with invasive fusariosis, especially those with high MIC of amphotericin B and voriconazole for FSSC, as well as other mould infections.³³

We found olorofim, another novel antifungal agent, showed limited in vitro activities against variable Fusarium SCs, except FFSC by measured at 90% inhibition of growth. Previous studies by the same reading endpoints have shown consistent high olorofim MICs against FSSC, FDSC and FIESC.^11,12,32 While these studies found differences in susceptibility at species level within FFSC and FOSC, our results showed uniformly low olorofim MICs against FFSC and high MICs against FOSC. In addition to relatively small numbers of FFSC (n = 8) and FOSC (n = 4) in our study, species difference with a specific SC may contribute to the discordant results. For example, one study by the EUCAST method found olorofim GMs of MICs for F. jonfreemaniae and F. musae was greater than that of F. annulatum and F. verticillioides (0.445 versus 0.124 versus 0.018 versus 0.081 mg/L) within 41 FFSC isolates.³² In our cohort, among eight FFSC isolates, only one F. annulatum was identified and showed a comparable olorofim MIC of 0.015 mg/L, while the other species differed from those in that study. These findings suggested that species-level identification within Fusarium SC is crucial for predicting antifungal susceptibility.

Regarding amphotericin B and voriconazole, two preferred agents for initial therapy of invasive fusariosis, this study found that FSSC displays higher MICs compared with non-FSSC, which is consistent with previous studies.^7,14,24,34 Overall, in vitro activities of voriconazole were very poor against all SCs, with modal MICs ≥4 mg/L and all MIC₉₀ > 16 mg/L. Furthermore, the distribution of amphotericin B MICs ranged from 0.125 µg/mL to more than 16 µg/mL for FSSC. By applying two-step molecular identification, we were able to demonstrate this variation is not random, but rather species-dependent within FSSC. Of note, N. keratoplastica, the second most common species of FSSC, exhibited higher amphotericin modal MICs (4 µg/mL) compared with other Neocosmospora species (0.5–1 mg/L), consistent to previous reports.¹⁴ While the proportion of non-wild type (WT) to amphotericin B (16 µg/mL or more) was highest for N. pseudensiformis (18%), the most common species of FSSC. Because commercial antifungal susceptibility tests for Fusarium have not yet been verified, molecular identification of Fusarium, especially FSSC, in conjunction with local susceptibility data may assist clinicians in selecting appropriate antifungal agents and/or titrate the optimal dose of amphotericin B.

The current study has several limitations. First, 70.8% of our clinical isolates were FSSC. Therefore, caution is advised when applying the two-step molecular identification and interpreting antifungal susceptibility results for non-FSSC or non-clinical isolates. In Netherland, a study focusing on FFSC also demonstrated antifungal resistance patterns are species specific.³⁵ Hence, species-level identification among non-FSSC may be practicable for the choice of antifungal treatment as well. Second, in vitro susceptibility testing for both novel and conventional antifungal agents were primarily conducted on FSSC with variable species. For accurate comparison with other studies, species-level identification within Fusarium SC is necessary, as species variations in susceptibility to a specific antifungal agent were observed. Third, regarding the clinical feasibility, we acknowledge that routine implementation of this two-step molecular strategy may be restricted by costs and technical capabilities. Therefore, we propose integrating this method into a tiered workflow, where molecular sequencing is reserved for isolates that fail species-level identification by MALDI-TOF MS. However, implementation of even this targeted approach remains challenging, as a recent survey indicated that DNA sequencing for fungal identification is available in only one-third of clinical laboratories in the Asia–Pacific region.³⁶ Hence, future multicentre studies are needed to validate these findings and, crucially, to use this molecularly characterized dataset to expand and refine MALDI-TOF MS reference databases, bridging the gap between high-resolution sequencing and rapid routine diagnostics. Fourth, our study lacked clinical data, which limits the ability to correlate antifungal MICs and clinical outcomes. Yet, previous research indicated that a clear correlation between in vitro activity and clinical effectiveness in invasive fusariosis may not exist.⁷

Collectively, the multicentre study demonstrates the utility of a two-step molecular identification protocol for accurately classifying Fusarium clinical isolates at the species level. These results refine understanding of the local epidemiology of blood and cornea Fusarium species, highlighting the predominance of FSSC and potential geographic variations at both SC and individual species levels. Manogepix displayed outstanding in vitro activities against all Fusarium isolates at the lowest tested concentrations. In comparison, susceptibilities to olorofim and amphotericin B had variations by SCs and/or species. These findings illustrate the significance of considering both species and species complex when predicting antifungal susceptibility, and such an alternative method may facilitate streamlined selecting optimal antifungal agents in clinical settings.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the authors used ChatGPT to enhance grammar and refine the English language. After employing this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

Funding

This study was supported by the Ministry of Science and Technology, Taiwan (112-2314-B-002-143-MY3). The funding authorities played no role in data collection, analysis, or interpretation; study design; writing of the manuscript; or decision to submit for publication.

Transparency declarations

None to be declared.

Author contributions

Pao-Yu Chen: Conceptualization, Methodology, Data curation, Writing—review & editing, Writing—original draft, Formal analysis, Visualization. Chi-Jung Wu: Data curation. Un-In Wu: Data curation. Wang-Da Liu: Data curation. Yee-Chun Chen: Conceptualization, Methodology Writing—review & editing, Resources, Project administration, Formal analysis, Supervision.

Institutional review board statement

The study was approved by the Institutional Review Board of NTUH and NTUCC (IRB number 202 304 096RIND) and NCKUH (IRB numbers B-ER-101–342, B-ER-104–057 and B-ER-109–364).

Data availability

The DNA sequences of all isolates in this study were deposited in GenBank (http://www.ncbi.nlm.nih.gov/) under BioProject accession number: PRJDB37776.

Supplementary data

Supplementary Materials and Tables S1 and S2 are available as Supplementary data at JAC-AMR Online.