Leukemia (2010) 24, 2032–2038 & 2010 Macmillan Publishers Limited All rights reserved 0887-6924/10 www.nature.com/leu ORIGINAL ARTICLE Diagnosis of invasive fungal infections by a real-time panfungal PCR assay in immunocompromised pediatric patients C Landlinger1, S Preuner1, L Basˇkova´1, M van Grotel2, NG Hartwig3, M Dworzak4, G Mann4, A Attarbaschi4, L Kager4, C Peters4, S Matthes-Martin4, A Lawitschka4, MM van den Heuvel-Eibrink2 and T Lion1 1Division of Molecular Microbiology, Children’s Cancer Research Institute (CCRI), Vienna, Austria; 2Department of Pediatric Oncology/Hematology, Erasmus MC-Sophia Children’s Hospital, Rotterdam, The Netherlands; 3Department of Infectious Diseases, Erasmus MC-Sophia Children’s Hospital, Rotterdam, The Netherlands and 4Department of Hematology/Oncology/Stem Cell Transplant Unit, St Anna Children’s Hospital, Vienna, Austria Invasive fungal disease (IFD) is a life-threatening event in Timely detection of the fungal pathogens is a prerequisite for immunocompromised patients, and there is an urgent need for successful therapy and the clinical outcome in patients with reliable screening methods facilitating rapid and broad detec- IFD. In view of the changing epidemiology and the increasing tion of pathogenic fungi. We have established a two-reaction real-time PCR assay permitting highly sensitive detection variety of fungal genera and species observed in immunosup- of more than 80 fungal pathogens, covering a large spectrum pressed patients, there is a growing demand for broad-spectrum of moulds, yeasts and Zygomycetes. To assess the clinical diagnostic tests. Many of the published technical approaches potential of the assay, more than 600 consecutive specimens based on real-time PCR methods are rapid and sensitive, from 125 pediatric patients carrying a high risk of IFD were but their detection spectrum is mostly restricted to a variety analyzed. An excellent correlation between PCR positivity and of Candida13,14 or Aspergillus species,15,16 or representatives the presence of proven, probable or possible fungal infection 17–21 according to the European Organization for Research and of both genera. Diagnostic assays permitting the detection Treatment of Cancer criteria was demonstrated, as revealed of a wide range of fungal pathogens including fungal species of by the sensitivity of the assay of 96% (95% CI: 82–99%). The both common and emerging fungal genera are therefore negative predictive value of the panfungal PCR assay presented required. To address the urgent clinical need, we have was 98% (95% CI: 90–100%), while the specificity and the developed a panfungal real-time PCR assay permitting the positive predictive value were 77% (95% CI: 66–85%) and 62% detection of at least 80 pathogenic fungal species (European (95% CI: 47–75%), respectively. The results indicate that molecular screening of patients during febrile neutropenic patent application 06817468.9). Herein, we describe the episodes by the assay presented could help prevent unneces- technical features of the assay, and present a combined sary toxicity resulting from empirical antifungal treatment prospective and retrospective study in severely immunocom- in individuals who may not be at risk of imminent fungal promised pediatric patients carrying a high risk of invasive disease. Our observations raise the possibility that rapid fungal infection, with the aim to assess the potential clinical species identification may be required to increase the positive benefit of molecular screening by a highly sensitive broad- predictive value for impending fungus-related disease. Leukemia (2010) 24, 2032–2038; doi:10.1038/leu.2010.209; spectrum fungus detection test. published online 30 September 2010 Keywords: invasive fungal infection; real-time PCR; panfungal PCR Materials and methods Fungal strains, bacterial and viral isolates Fungal strains for PCR testing (Table 1) were obtained from Introduction different institutions, including the American Type Culture Collection (ATCC, Rockville, USA), the German Collection of Invasive fungal disease (IFD) is a leading cause of morbidity and Micro-organisms (DSM, Braunschweig, Germany) and the mortality in severely immunocompromised individuals, includ- Institute of Hygiene and Medical Microbiology of the University ing particularly patients with hematological malignancies of Vienna (IHMM, Austria). and bone-marrow transplant recipients. The vast majority of IFD events are still caused by Candida and Aspergillus species; Clinical materials however, changes in the epidemiology have occurred over the Clinical specimens from consecutive patients were obtained past decades.1–5 Recent studies from North American and after informed consent, and were prospectively collected as European centers indicate an increasing incidence of previously specified below. In total, 618 peripheral blood specimens from uncommon fungal genera, such as Cryptococcus, Trichosporon, 125 pediatric hemato-oncological patients undergoing intensive Malassezia,6,7 Fusarium,8,9 Scedosporium, Pseudallescheria,10,11 chemotherapy (n ¼ 65) or allogeneic stem cell transplantation and, very importantly, different members of the class Zygomycetes, (n ¼ 60) were analyzed during 150 episodes of febrile neutro- such as Rhizopus and Mucor.12 penia. Whenever possible, specimens were collected at first onset of fever, within 48 h thereafter, and at subsequent time Correspondence: Professor T Lion, Division of Molecular Microbiology, points in the course of the febrile episode, upon availability. On Children’s Cancer Research Institute (CCRI), Zimmermannplatz 10, average, four peripheral blood samples were investigated during Vienna A-1090, Austria. E-mail: [email protected] each episode (range 1–15). In selected instances, depending Received 15 March 2010; revised 16 July 2010; accepted 29 July on availability, additional types of specimens derived from 2010; published online 30 September 2010 primarily sterile sites of suspected infection were collected and Panfungal real-time PCR assay for diagnosis of IFD C Landlinger et al 2033 Table 1 Fungal species/isolates detected by the panfungal Table 1 (Continued ) real-time PCR assay Trichosporon asahii Reaction I Trichosporon cutaneum Experimentally tested Trichosporon inkin Acremonium strictum Saccharomyces cerevisiae Alternaria alternata Aspergillus candidus Based on sequence alignments Aspergillus clavatus Apophysomyces elegans Aspergillus flavus Basidiobolus ranarum Aspergillus fumigatus Cokeromyces recurvatus Aspergillus glaucus Mucor circinelloides Aspergillus nidulans Mucor ramosissimus Aspergillus niger Mucor rouxii Aspergillus terreus Rhizomucor variabilis Aspergillus versicolor Rhizopus azygosporus Fusarium oxysporum Rhizopus microsporus Fusarium proliferatum Rhizopus stolonifer Fusarium solani Rhodotorula mucilaginosa Fusarium verticillioides Saksenaea vasiformis Penicillium chrysogenum Trichosporon asteroides Penicillium citrinum Trichosporon moniliiforme Penicillium marneffei Trichosporon mucoides Penicillium purpurogenum Trichosporon ovoides Penicillium simplicissmum Scedosporium apiospermum Human pathogenic fungi covered by the panfungal PCR detection assay including reactions I and II. Based on sequence alignments Aspergillus ochraceus Aspergillus penicillioides Aspergillus ustus Bipolaris eleusines Cladosporium cladosporioides subjected to molecular analysis. These included cerebrospinal Cladosporium oxysporum fluid (n ¼ 11) and lung biopsies (n ¼ 2) in patients with suspected Histoplasma capsulatum central nervous system or pulmonary involvement, respectively, Scedosporium prolificans in order to correlate the data with the PCR findings in peripheral blood samples. The specimens used as training set for initial data Reaction II Experimentally tested assessment were provided by the Erasmus MC-Sophia Children’s Absidia corymbifera Hospital, Rotterdam, The Netherlands, and the specimens used Blastoschizomyces capitatus as validation set were provided by the St Anna Children’s Candida albicans Hospital of Vienna, Austria. Plasma or serum was usually Candida allociferrii collected from 3-ml samples of peripheral blood and all samples Candida colliculosa were stored at À80 1C until further processing. Peripheral blood Candida cylindracea Candida dubliniensis specimens from healthy volunteer donors were used to test for Candida famata cross-reactivity with human DNA (see below). Candida glabrata Candida guilliermondii Candida inconspicua DNA extraction Candida kefyr The isolation and purification of fungal DNA were performed Candida krusei 17 Candida lambica essentially as described earlier, with minor modifications. All Candida lipolytica steps were performed in a laminar flow hood using one-way Candida lusitaniae sterile utensils and sterile-filtered reagents. DNA isolation Candida membranaefaciens from bacteria and virus specimens for the assessment of cross- Candida norvegensis reactivity was performed using the QIAamp DNA Mini Kit Candida parapsilosis (QIAGEN GmbH, Hilden, Germany). Candida pelliculosa Candida rugosa Candida sake Candida tropicalis Real-time PCR Candida utilis The panfungal PCR detection assay included two separate Candida zeylanoides reactions, I and II, each covering different subsets of fungal Cryptococcus albidus pathogens, as outlined in Table 1. Both PCR reactions targeted a Cryptococcus laurentii Cryptococcus neoformans highly conserved region of the 28S ribosomal RNA multicopy Cunninghamella bertholletiae gene. The PCR reactions were set up in a total volume of