Prevalence of fungal DNAemia mediated by putatively non-pathogenic fungi in immunocompromised patients with febrile neutropenia: a prospective cohort study

Abstract

Invasive fungal disease (IFD) presents a life-threatening condition in immunocompromised patients, thus often prompting empirical administration of antifungal treatment, without adequate mycological evidence. Over the past years, wide use of antifungal prophylaxis resulted in decreased occurrence of IFD but has contributed to changes in the spectrum of fungal pathogens, revealing the occurrence of previously rare fungal genera causing breakthrough infections. The expanding spectrum of clinically relevant fungal pathogens required the implementation of screening approaches permitting broad rather than targeted fungus detection to support timely onset of pre-emptive antifungal treatment. To address this diagnostically important aspect in a prospective setting, we analyzed 935 serial peripheral blood (PB) samples from 195 pediatric and adult patients at high risk for IFD, involving individuals displaying febrile neutropenia during treatment of hematological malignancies or following allogeneic hematopoietic stem cell transplantation. Two different panfungal-PCR-screening methods combined with ensuing fungal genus identification by Sanger sequencing were employed. In the great majority of PB-specimens displaying fungal DNAemia, the findings were transient and revealed fungi commonly regarded as non-pathogenic or rarely pathogenic even in the highly immunocompromised patient setting. Hence, to adequately exploit the diagnostic potential of panfungal-PCR approaches for detecting IFD, particularly if caused by hitherto rarely observed fungal pathogens, it is necessary to confirm the findings by repeated testing and to identify the fungal genus present by ensuing analysis. If applied appropriately, panfungal-PCR-screening can help prevent unnecessary empirical therapy, and conversely, contribute to timely employment of effective pre-emptive antifungal treatment strategies.

Trial registration: ClinicalTrials.gov NCT02492594.

Keywords: Antifungal therapy; Antifungal treatment; Fungal diagnostic; Invasive fungal disease; panfungal-PCR.

Fig. 1

Most commonly detected fungal genera in the cohorts studied and the presumptive pathogenicity of the observed fungal spectrum in the immunocompromised setting. Panel A: Overall, almost 85% of the PB samples tested negative by both panfungal-PCR approaches employed. The most commonly observed fungal genus was Cladosporium, which was detected in 2.7% of all PB samples tested. Panels B and C: The fungal genera detected in pediatric and adult patients revealed no statistically significant differences between the two cohorts, although the genus Malassezia was slightly more common in the pediatric setting (2.6% of all PB samples tested in the pediatric versus 0.9% in the adult cohort). Panels D and E: The distribution of fungal genera detected in patients with IFD at any level and in patients with absent IFD was very similar. However, the diversity of fungal genera was smaller in the subset of patients with IFD at any level, which could be attributable to the smaller sample size. Panel F: Overall, over 50% of the fungal genera identified in PB samples from high-risk patients using panfungal screening and sequencing of ITS2-PCR amplicons were classified as putatively non-pathogenic (indicated as absent pathogenicity in the Figure), when applying the search parameters outlined in the Supplementary Material. The frequency of fungi displaying proven, probable, possible, or absent pathogenicity revealed no statistically significant differences between the pediatric and adult cohorts studied (the numbers for fungal genera with proven pathogenicity were 8.0% vs. 13.6%, p = 0.3; with probable pathogenicity 12.0% vs. 22.7%, p = 0.1; with possible pathogenicity 28.0% vs. 27.3%, p = 0.7; with absent pathogenicity 52.0% vs. 36.4% p = 0.6). n, number of PB specimens investigated. Panel G: Fungi displaying possible, probable, or proven pathogenicity according to the search parameters applied were observed in 24.6%, 11.5%, and 9.8% of PB samples, when the presence or absence of IFD was not considered (all). However, between patients displaying IFD at any level and patients with absent IFD, there was a statistically significant difference in the frequency of fungal genera with probable pathogenicity (38.5% vs. 12.3%, p = 0.02) and absent pathogenicity (15.4% vs. 56.1%; p = 0.008). Importantly, however, the fungal DNAemia detected was only transient and could not be confirmed in sequential analyses