We conducted this study in 18 hospitals in Spain, 8 of which belong to the Spanish Network for Research in Infectious Diseases, Instituto de Salud Carlos III, in Madrid, Spain. The study was approved by Reina Sofía University Hospital Institutional Review Board (Córdoba, Spain), which waived the need to obtain written informed consent. During January 2000–December 2015, hospital staff reviewed microbiological and pathologic registries to identify cases of IF. Data were recorded in a password-protected, electronic clinical research file. We monitored the collected data for missing information, inconsistencies, and ambiguities; when necessary, we sent queries to the appropriate hospitals for clarification.

We conducted a blind review of reported cases of IF. In this study, we included only proven cases of IF according to the consensus definitions of invasive fungal diseases established by the European Organization for Research and Treatment of Cancer and the Mycoses Study Group Education and Research Consortium (16,17). IF can be proven in 4 ways: microscopic examination of a specimen obtained by needle aspiration or biopsy that documents hyphae and isolates Fusarium spp. in the same tissue; blood culture yielding Fusarium spp. alongside signs consistent with an infectious disease process; isolation of Fusarium spp. in a normally sterile site (excluding bronchoalveolar lavage fluid, a paranasal or mastoid sinus cavity specimen, and urine) with accompanying signs of infection; or amplification and sequencing of Fusarium DNA in formalin-fixed paraffin-embedded tissue (16,17).

The participating hospitals identified Fusarium isolates using genetic sequencing, matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry, or morphologic characteristics. In addition, all biological samples available at the time of this retrospective study (i.e., Fusarium isolates and specimens for biopsies) were sent to the Mycology Reference Laboratory at the Instituto de Salud Carlos III for confirmatory genetic sequencing. The isolates were freeze-dried on potato agar slants on arrival and stored in distilled water. Molecular identification was based on the translation elongation factor 1α gene (18). Isolates were cultured in GYEP medium (0.3% yeast extract, 1.0% peptone; Francisco Soria Melguizo S.A., http://www.f-soria.es) with 2.0% glucose (Sigma-Aldrich Inc., https://www.sigmaaldrich.com) at 30°C for 24–48 h. Genomic DNA was isolated by using an extraction procedure described previously (19). Internal transcribed spacer region and a portion of the translation elongation factor 1α gene region were amplified as previously described (20). Alignment results were obtained and edited using Lasergene MegAlign Pro software from DNASTAR, Inc. (https://www.dnastar.com). All sequences were compared with reference sequences from GenBank and MycoBank (https://www.mycobank.org) databases by using InfoQuest FP version 4.50 software (Bio-Rad Laboratories, https://www.bio-rad.com). We also used an in-house database of the Mycology Reference Laboratory. When only biopsy specimens were available, we used real-time PCR specific for Fusarium species as previously described (21). When identified cases did not have available isolates or biologic material for PCR, we reported the specific Fusarium species only if obtained by MALDI-TOF mass spectrometry in conjunction with histopathology. We reported the remaining IF cases in patients with compatible clinical signs and symptoms as caused by Fusarium spp.

We defined survival as the time between the date of diagnosis and death or end of follow-up care (i.e., 90 days). We defined the date of diagnosis as the day of the first Fusarium-positive culture. We defined disseminated fusariosis as the involvement of >1 noncontiguous organ. We did not consider fungemia cases as disseminated disease unless >1 organ was involved (e.g., skin, lung, or sinuses) (3). We defined neutropenia as a blood neutrophil count <500 cells/mm³ temporally related to the onset of fungal disease. We defined persistent neutropenia as cases with continued low neutrophil counts at the end of treatment or death. We defined resolution of neutropenia as a persistent recovery of blood neutrophil count >500/mm³ as determined by available hospital records.

We calculated the incidence of IF using total admissions in the participating hospitals as the denominator; we expressed the incidence as the number of cases per 100,000 admissions. For statistical purposes, we arbitrarily defined 2 time periods: 2000–2009 and 2010–2015. We compared incidences between the different periods by χ² test using WinPepi version 11.65 (Brixton Health, http://www.brixtonhealth.com). We considered p<0.05 to be significant.

We collected variables describing patient demographic data, concurrent conditions, neutropenia, receipt of corticosteroids, clinical manifestations, diagnostic procedures, treatment, and outcome (i.e., 90-day survival). We compared categorical variables using χ² or 2-tailed Fisher exact test, and we compared continuous variables using the Mann-Whitney U test. We conducted all statistical tests with SPSS Statistics 16.0 (IBM Inc., https://www.ibm.com) and R version 3.5.0 (R Foundation for Statistical Computing, https://www.r-project.org); we considered 2-sided p values <0.05 to be significant. We constructed unadjusted Kaplan–Meier curves and compared them using the log-rank test. In addition, we evaluated factors associated with 90-day survival. We evaluated the following variables by univariate and multivariable Cox regression analyses: age, sex, age-adjusted Charlson comorbidity index (22), underlying disease, receipt of antifungals and corticosteroids in the previous 30 days, neutropenia at diagnosis of fusariosis and at the end of follow-up, clinical manifestations, and primary treatment (monotherapy with a lipid formulation of amphotericin B, monotherapy with voriconazole, or combination treatment). We entered variables with p<0.1 by univariate analysis into the multivariate analysis; we included variables with p<0.05 by the multivariate analysis in the final model. We evaluated the prediction accuracy of the final Cox model using the area under the receiver operating characteristic curve.

We described 58 cases of invasive Fusarium infections at 18 hospitals in Spain during a 15-year period (2000–2015). During this time, IF incidence increased from 0.40 (2000–2009) to 0.79 (2010–2015) cases per 100.000 admissions (p<0.01). Incidence of IF in neutropenic patients increased from 0.32 to 0.57 cases per 100,000 admissions (p = 0.06). We observed a 3-fold increase in the incidence of IF in nonneutropenic patients, from 0.08 to 0.22 cases per 100,000 admissions (p = 0.05). This increase might have been caused by an increase in the at-risk population, environmental exposure to Fusarium conidia, the increased use of antifungal prophylaxis, or a combination of these factors.

To date, the highest IF incidence rates worldwide are in healthcare centers in Brazil. A cohort of HSCT recipients at hospitals in Brazil during 1985–2001 had a 0.6% prevalence of IF (1). During 2007–2009, ≈10 years later, a prospective multicenter study by Nucci et al. reported a 3.7% IF prevalence among allogeneic HSCT patients and 3.4% among patients with acute myeloid leukemia (7). That study also found a 1-year cumulative incidence of 5.2% among those who had received an allogeneic HSCT and 3.8% among those with acute myeloid leukemia or myelodysplasia (7). A >10-fold increase, from 86 to 1,023 IF cases per 100,000 admissions, occurred among patients with hematologic malignancies at a healthcare center in Brazil from 2000–2005 to 2006–2010 (5). In the United States, a prospective multicenter study conducted by the Transplant-Associated Infection Surveillance Network during 2001–2006 reported a 1-year cumulative incidence of up to 0.3% of non-Aspergillus mold infections (23). In Italy, a multicenter retrospective study reported a 0.2% incidence of IF among patients who had received an allogenic HSCT during 1999–2003 (24).

In this cohort in Spain, IF occurred frequently in patients with hematologic conditions and in HSCT recipients. IF occurs almost entirely in markedly immunosuppressed patients who usually are neutropenic; we identified IF in nonneutropenic patients, including patients with a history of solid organ transplants, chronic cardiac or lung disease, or rheumatoid arthritis. These findings are in agreement with Park et al. (9), who studied 37 patients with IF and noted that 54.1% of IF patients were nonneutropenic, 83.8% had hematologic malignancies, and 16.2% had history of solid organ transplantation (9). Thus, nonneutropenia and certain nonhematologic conditions might not be uncommon among IF patients.

Lungs, the bloodstream, and the skin were the organs most frequently affected by IF. Fusarium species produce aleuroconidia, yeastlike structures that can invade the bloodstream and might cause fungemia and metastatic skin lesions (25).

IF is associated with high death rates. This cohort had a 90-day death rate of 56.9%, similar to rates noted in other studies (3,4,12,26). The death rates of nonneutropenic and neutropenic patients who recovered from neutropenia were similar (28.6% of nonneutropenic patients vs. 38.1% of neutropenic patients; p = 0.56), consistent with previous reports that 70% of IF cases resolve when patients recover from neutropenia (27). In this cohort, patients with persistent neutropenia had a 91.3% death rate despite antifungal therapy. Persistent neutropenia was the single most predictive prognostic factor in IF, consistent with previous reports (2).

Researchers must identify effective therapeutic strategies to improve the prognosis of IF patients with persistent neutropenia. The treatment practices observed in our study align with guidelines from the European Society of Clinical Microbiology and Infectious Diseases Fungal Infection Study Group and European Confederation of Medical Mycology (13). We examined the effects of different therapeutic regimens on 90-day death rate. We found no differences in the outcome of patients treated with voriconazole and those receiving a lipid formulation of amphotericin B. Combination therapy also was not associated with outcome, consistent with results from the largest study series conducted so far (3,12). However, these results must be interpreted with caution. Nucci et al. (3) found that patients receiving combination therapy had more severe disease (3), possibly influencing the clinician’s decision to administer 2 drugs. In our study, no nonneutropenic patients received combination therapy; further studies should examine the potential benefits of combination therapy in nonneutropenic patients. Finally, because we observed similar clinical responses in patients treated with a lipid formulation of amphotericin B or voriconazole as monotherapies, clinicians might not need to rely on antifungal susceptibility tests to guide treatment. Most clinically relevant Fusarium isolates exhibit high minimal inhibitory concentrations to most antifungals, including azoles, echinocandins, and polyenes (28–30).

Our study is subject to the limitations of retrospective observational studies. For example, the sample size was limited to the cases with available data. Despite being a multicenter study during a 15-year period, the sample size might be insufficient to detect significant differences in some groups. Furthermore, clinicians might prescribe a more potent treatment regimen for patients with severe disease, possibly skewing our analysis of treatment outcomes. Although all IF cases in this retrospective study met standard criteria (16,17), we could not determine the causative species in every case. Thus, the Fusarium species was only reported when determined by genomic sequencing or MALDI-TOF mass spectrometry, because these techniques have high agreement rates (89.8%–97.0%) (31–33). All remaining IF cases in our series were reported as Fusarium spp. We also were not able to culture all Fusarium isolates because of a lack of specimens, especially for cases occurring before 2010 in hospitals in which Fusarium isolates were not collected as part of routine procedures

In conclusion, our data show that IF is an emerging infection in Spain. We report an increase in the incidence of IF among nonneutropenic patients, including those with hematologic conditions and other concurrent conditions, such as chronic cardiac or lung diseases, rheumatoid arthritis, or history of solid organ transplants. Our results support previous studies reporting that IF survival is critically dependent on resolution of neutropenia.