Fifty isolates from clinical or environmental sources, morphologically identified as F. solani (6), were included in the study (Table 1). Isolates were stored by lyophilization and submerged in slant cultures in mineral oil at room temperature. The procedures for DNA extraction, amplification, and sequencing of the different regions analyzed were described by Gilgado et al. (8). The annealing temperature was 55°C, and the primers used were EF-1H and EF-2T (16) for EF-1α, TUB-F (5) and T22 (17) for β-tubulin, and ITS5 and ITS4 (28) for the ITS. The phylogenetic analysis was performed using PAUP*, version 4.0b10 (27). Maximum parsimony trees were obtained after 100 heuristic searches with random sequence addition and tree bisection-reconnection branch-swapping algorithms, collapsing zero-length branches, and saving of all minimal-length trees (MulTrees). The results of the partition homogeneity test showed that the three locus sequence data sets were congruent (P = 0.2) and could therefore be combined. Sequences of the three genes were analyzed phylogenetically as separate (data not shown) and combined data sets.

TABLE 1. Isolates included in this study

We evaluated the in vitro activities of 11 antifungal drugs against 27 representative strains (22 clinical and 5 environmental) randomly chosen from the main clades obtained in the phylogenetic analysis. The isolates were grown on potato dextrose agar plates and incubated at 25°C for 7 days. We used a microdilution reference method (14), with some modification. The inocula were adjusted to a final concentration of 4 × 10³ to 5 × 10⁴ conidia/ml with a hemocytometer and verified by quantitative colony counts on potato dextrose agar plates. Paecilomyces variotii ATCC 36257 was included in each batch of tests as a quality control strain. The antifungal agents tested were AMB, albaconazole, VRC, itraconazole, ravuconazole, terbinafine, ketoconazole (KTC), posaconazole (PSC), micafungin (MFG), fluconazole (FLC), and flucytosine (5-FC). MFG, FLC, and 5-FC were diluted in sterile distilled water, and the rest were diluted in dimethyl sulfoxide. Final drug concentrations ranged from 64 to 0.12 μg/ml for FLC and 5-FC, from 128 to 0.25 μg/ml for MFG, and from 16 to 0.03 μg/ml for the rest. The MIC end point for the triazoles and AMB was defined as the lowest concentration that produced complete inhibition of growth, and that for FLC, KTC, 5-FC, and MFG was defined as the lowest concentration that produced a 50% inhibition of growth.

With the primers used, we were able to amplify and sequence 654, 461, and 573 bp of the EF-1α, β-tubulin, and ITS gene sequences, respectively. Parsimony analysis of the combined data set (1,688 bp) yielded three trees of 176 steps in length. Four main clades (I, II, III, and IV) with high bootstrap support were obtained, resulting in a total of 28 different haplotypes and numerous putative cryptic species (Fig. 1). We found no relationship between the biogeographical origins of the isolates and the molecular groups. In order to compare the topology of our trees with those obtained by Zhang et al. (29), we performed a new phylogenetic analysis based on ITS and EF-1α loci (data not shown). We included our isolates and several representative sequences, retrieved from GenBank, from Zhang et al.'s main groups. In general, our clades did not coincide with those of Zhang et al., which confirms the high genetic variability of this complex. However, several of our strains nested in their groups 3 and 4.

FIG. 1. One of the three most parsimonious trees obtained from heuristic searches, based on a combined data set. Bootstrap support values above 70% are indicated at the nodes. CI, consistency index; RI, retention index; HI, homoplasy index.

Results of the in vitro susceptibility tests are shown in Table 2. In general, all drugs showed high MICs, with no remarkable differences among the clades. AMB was the most active drug, although in no case was the MIC lower than 1 μg/ml, followed by VRC. The latter drug showed the most variable results; sometimes the differences among MICs against different isolates in the same clade were >8-fold.

TABLE 2. Distribution of 27 tested isolates from the four phylogenetic clades of F. solani, according to antifungal susceptibility

The poor in vitro activities of different antifungal drugs against Fusarium have been reported by other authors many times, with F. solani being more resistant than the other species of the genus (12, 19, 24, 26). In this study, AMB was revealed as the most active drug, but only one isolate from clade I and another from clade III showed a MIC of 1 μg/ml, with all others showing higher values. Although this drug is recommended for the treatment of fusariosis, it has poor clinical success (7, 11, 15). In cases involving neutropenic patients receiving corticosteroids, the survival rate is practically nil, despite aggressive treatment (15). The poor efficacy of AMB was also demonstrated in animal studies (10). VRC, the other recommended drug, is the only agent indicated for treating refractory fusariosis (7). However, in our study, the VRC MICs were always higher than 2 μg/ml, which agrees with the results of other studies (12, 26). This drug was effective in a few clinical cases of fusariosis (2, 4), although in none of them was the species involved referred to as F. solani. This is probably linked to the fact that in animal studies F. solani was clearly more virulent and more difficult to treat than F. oxysporum and F. verticillioides, the two other common species of the genus (13, 18). In a recent clinical trial, 45.5% (5/11) of patients responded satisfactorily to VRC (20), although unfortunately the isolates were not identified to the species level. Arikan et al. (1) demonstrated that the combination of caspofungin plus AMB was synergistic against 100% of strains of F. oxysporum but against only 25% of F. solani strains. In the case of PSC, we also obtained high MICs, again agreeing with other studies (19, 26). However, in a recent retrospective analysis, PSC as an aggressive treatment for invasive fusariosis gave a 48% successful outcome (25), but the species involved were not listed. In conclusion, the F. solani species complex constitutes a group of genetically diverse fungi with poor in vitro and in vivo responses to different antifungal drugs.

We thank Félix Gilgado, Carol Serena, Rita Marimon, and Marçal Mariné for their contributions to this work.