Revision and Update of the Consensus Definitions of Invasive Fungal Disease From the European Organization for Research and Treatment of Cancer and the Mycoses Study Group Education and Research Consortium

1(See the Editorial Commentary by Rogers on pages 1377–8.)

2The European Organization for Research and Treatment of Cancer and the Mycoses Study Group Education and Research Consortium (EORTC/MSGERC) consensus definitions of invasive fungal diseases (IFDs) were last updated in 2008. These definitions achieved their original aim in fostering communication and enabling comparison of study findings among those engaged in research into IFD of patients with cancer and recipients of hematopoietic stem cell transplants (HSCTs) or solid organ transplants (SOTs). Moreover, they have been adopted by regulatory agencies for evaluation of antifungals and have been used to evaluate diagnostic tests and to conduct epidemiologic studies. Importantly, these definitions are specifically intended for these purposes only and not to direct or guide patient care.

3The 2008 definitions had their shortcomings, including the facts that the definitions were unsuitable for patients with IFD in the setting of intensive care units (ICUs) or in pediatrics, data were insufficient to establish appropriate thresholds for detecting Aspergillus galactomannan (GM), and there was uncertainty about the role of (1,3)-beta-D glucan (BDG). Furthermore, nucleic acid amplification including polymerase chain reaction (PCR)–based tests were excluded because of lack of standardization and validation. Definitions for cryptococcosis and endemic mycoses also needed clarification, and there were no definitions for pneumocystosis.

4Volunteers from the EORTC Infectious Diseases Group and the MSGERC were assigned according to their expertise to 10 working groups, each charged with appraising a particular topic (see list of contributors in the Notes section). The chairs of the EORTC and MSG (J. P. D. and P. G. P.) selected leaders for each working group, and S. C. served as executive secretary. After completing the first round of working group assignments, leaders presented each group’s initial deliberations and recommendations at the 7th Trends in Medical Mycology Conference in Lisbon, Portugal, October 2015. A slide set was made available until 31 December 2015 online at www.e-materials.com/timm2015/invitation/Member and, on request, for public comment. After several iterations, the final draft of the manuscript was circulated to all members for their approval.

5Pediatrics and patients in the ICU were considered as special populations. However, group 10 (IFD definitions in ICU patients) was unable to generate recommendations that preserved a level of certainty consistent with the existing definitions except for proven IFD (Table 1) and therefore undertook a separate initiative.

6There was a clear need to establish pediatric-specific IFD definitions as the clinical and radiologic manifestations of IFD in children, that is, patients aged <18 years, may differ significantly from those in adults. Most importantly, the incidence of invasive candidiasis (IC) is higher in neonates than in other age groups. The degree of prematurity, based on gestational age and birth weight, is a unique risk factor among neonates; hematogenous Candida meningoencephalitis affects premature infants disproportionately and has serious consequences including seizures, intraventricular hemorrhage, and developmental delay. With respect to IC due to non-Candida albicans species, Candida glabrata is the most common pathogen in adults, whereas Candida parapsilosis predominates in children and neonates. Risk factors for invasive mold diseases include innate immunologic defects, with Aspergillus nidulans being associated with chronic granulomatous disease, while Aspergillus fumigatus is seen more often in other patient groups.

7Neonates with IC often present with subtle clinical findings, and cultures are frequently sterile, including cerebrospinal fluid (CSF) samples from neonates with candidemia and concurrent Candida meningitis. Diagnosis is often inferred from insensitive and nonspecific surrogate tests, such as increased C-reactive protein or thrombocytopenia, which has been shown to be a predictor of candidemia in infants. In neonates, a positive urine culture has a significance similar to that of a positive blood culture as an indicator of IC. Radiographic findings are less specific in children than those reported in adults. Chest computerized tomography (CT) scans in children with proven invasive pulmonary aspergillosis (IPA) commonly show nonspecific changes and not the halo sign, air crescent formation, or cavitation seen in adults.

8There are also far fewer data to support the clinical use of nonculture-based fungal biomarkers in neonates and children, although the GM assay performs similarly in children and adults when used as an adjunctive tool to diagnose invasive aspergillosis (IA). Likewise, there are few data regarding the use of BDG, Candida mannan antigen, and anti-mannan antibody biomarkers in pediatrics. Recent data support the utility of BDG in CSF for the diagnosis and therapeutic monitoring of children with Candida meningoencephalitis, but the data are sparse regarding the utility of PCR assays and the T2Candida assay for diagnosis.

9In the previous definitions, indirect tests for diagnosing IFD were only included if there was sufficient evidence that they had been standardized and validated. Moreover, commercial tests were included only if criteria for interpretation had been provided. Hence, while tests for GM and BDG were incorporated, tests for detecting fungal nucleic acid were not. Furthermore, there was no agreement about appropriate thresholds, so the manufacturers’ analytical thresholds were adopted. The evidence for using GM to diagnose IA has grown considerably since then, and testing for BDG has been extended to a wide range of patients. With respect to Aspergillus PCR, the International Society of Human and Animal Mycology working group Fungal PCR Initiative (FPCRI; www.fpcri.eu) has made significant progress toward setting a standard for the technique after vigorous validation.

10There is mounting evidence that the radiologic manifestations of invasive mold disease are more varied than previously recognized. The increased sensitivity of newer imaging techniques enables a greater number and depth of abnormalities to be seen in different anatomic regions. Recent data relating to the role of imaging in the diagnosis of IPA and pulmonary mucormycosis (PM) in adults suggest that a high-resolution CT scan (HRCT) is preferred to chest radiographs, magnetic resonance imaging (MRI), and positron emission tomography (PET), likely reflecting that HRCT is more sensitive than a chest radiograph, more widely available than MRI, and the experience with HRCT is much larger than with PET. Among patients with IPA, nodules or infiltrates with a halo sign remain useful among neutropenic patients but they are nonspecific for IPA in other groups. Furthermore, the air crescent sign is a late and nonspecific sign. Among nonneutropenic patients, multiple pulmonary nodules and various nonspecific findings including bronchopneumonia, consolidation, cavitation, pleural effusions, ground glass opacities, tree-in-bud opacities, and atelectasis are found. Overall, consolidation is the most frequent presentation of PM, followed by mass lesions, nodules, and cavitation. Multiple nodules (more than10) and pleural effusions appear to be more frequent in PM than in IPA. Moreover, the reverse halo sign is more specific for PM than IPA, although the differential diagnosis also includes other diseases including tuberculosis.

11We evaluated Aspergillus galactomannan for both adults and children and specific patient groups and its utility and validity for different clinical specimens. We adopted different thresholds for different specimens rather than for different host groups (Table 2). These differ from those recommended by the manufacturer of the GM assay (Platelia Aspergillus (Bio-Rad, CA), validated only for use in serum and bronchoalveolar lavage (BAL) fluid; however, detection of GM in plasma and CSF should support a diagnosis of IA. Exposure to mold-active antifungals compromises the utility of the GM test for IA by reducing its sensitivity. Therefore, caution should be exercised when GM is found to be absent from serum or plasma in patients receiving mold-active antifungals. There was consensus that similar GM thresholds are appropriate for adults and children.

12The group considers detection of BDG to be suitable for diagnosing probable IFD in the appropriate clinical setting. This includes patients with hematologic malignancies with and without neutropenia, neutropenia following HSCT, and certain patients in the ICU who are at higher risk (>10%) for IC as a result of gastrointestinal surgery with recurrent anastomotic leaks, perforations of the upper gastrointestinal tract, or necrotizing pancreatitis when there is clinical suspicion of infection. A single threshold (>80 pg/mL) using the Fungitell test (Associates of Cape Cod, Falmouth, MA) is recommended; there is insufficient evidence to include assays produced by other manufacturers. Confidence for true positive results increases with repeated positive tests and for values that greatly exceed the positivity threshold. There may be variability in positive predictive value (PPV) and negative predictive value (NPV) based on patient population, but a single threshold is favored at this time. The group did not support the use of serum BDG to rule in patients for clinical trials or for defining IA or IC, as BDG detection is not specific for any one IFD. It was agreed that this test should only be used on serum samples, although the test has been used for CSF samples with some success to support a diagnosis of central nervous system (CNS) IFD in certain circumstances when other diagnostic tests are negative or inconclusive.

13The T2Candida panel has been cleared by the US Food and Drug Administration for the detection of common Candida species from whole blood specimens. The test has a very high NPV but, as with all such tests, the PPV is variable and depends upon disease prevalence in a given patient population. The PPV increases from 62% among patients with sepsis, shock, or lengths of stay greater than 3–7 days in an ICU to 92% for bone marrow transplant recipients and patients with leukemia who are neutropenic but not receiving any antifungal prophylaxis. The test has been included as mycologic evidence to support a diagnosis of candidemia in selected clinical trials.

14In considering Aspergillus PCR, target species, patient populations, appropriate specimens for testing, technical issues, comparison with other biomarker assays, and unique attributes of PCR assays were reviewed. The data were sufficiently robust for performing Aspergillus PCR on serum, plasma, whole blood, and BAL fluid in adults. The group acknowledged that Aspergillus PCR data have been evaluated most extensively for adults with hematologic malignancies and HSCT. Systematic reviews of Aspergillus PCR methods on blood and BAL fluid conclude that PCR provides a robust diagnostic test for screening and confirming the diagnosis of Aspergillus infection.

15There are relatively few commercial PCR assays, and most investigators have developed methods in-house. As such, the FPCRI was established to develop criteria for Aspergillus PCR rather than a standardized method per se. Despite technologic variability, PCR performance was comparable with that for detecting GM and BDG. Moreover, commercial PCR tests performed using methodology in line with the FPCRI recommendations provide a standardized approach that has been independently associated with improved performance. A unique feature of PCR is its ability to detect both genus and species of Aspergillus. PCR is also capable of identifying certain mutations associated with triazole resistance directly from clinical specimens.

16Tissue diagnosis requires the presence of fungal elements in formalin-fixed paraffin-embedded tissue and signifies proven fungal disease but not the identity of the fungus involved. To achieve this, we recommend amplification of fungal DNA by PCR combined with DNA sequencing, but only when fungal elements are seen by histopathology. PCR would add value by allowing identification of the fungus to genus and possibly species levels. Because the technique used should be rigorously quality controlled, only laboratories with a proven record in performing DNA extraction from formalin-fixed tissue should undertake this. The identity of the fungus should be consistent with the histopathologic findings.

17The inclusion of Pneumocystis jirovecii pneumonia (PCP) diagnosis in the updated definitions was limited to patients not living with human immunodeficiency virus (HIV). Diagnosing PCP has been more difficult among these patients possibly due to a more focal pulmonary involvement, lower suspicion of disease, and lower sensitivity of traditional histologic and microscopy diagnostic tests. As such, it is important to more fully define host factors for patients at increased risk for PCP. We selected receipt of therapeutic doses of corticosteroids for at least 2 weeks within the past 60 days; antineoplastic, antiinflammatory, or immunosuppressive treatment; and low CD4 lymphocyte counts due to a medical condition. This includes, but is not limited to, patients with primary immunodeficiencies, hematologic malignancies, SOTs, and allogeneic HSCT recipients. Clinical criteria in this population tend to be nonspecific and include cough, dyspnea, and hypoxemia. Radiographic abnormalities include bilateral ground-glass opacities and, less frequently, consolidation, small nodules, unilateral infiltrates, pleural effusions, and cystic lesions. Amplification of P. jirovecii DNA by quantitative real-time PCR on BAL fluid, expectorated sputum, or oral wash specimens is preferred to qualitative PCR and is helpful to establish probable disease. However, further studies are needed to validate thresholds for positivity. Similarly, 2 or more serum BDG levels of ≥80 ng/L are useful for diagnosing probable disease in appropriate clinical context provided other IFDs have been excluded.

18A broader understanding of the natural history and host factors associated with cryptococcal disease warrants updating these definitions. We support the previous definitions of proven and probable cryptococcal disease in any host. However, we also recognize cryptococcal infection among individuals in high-risk host groups who have few, if any, symptoms and only a positive serum cryptococcal antigen test (asymptomatic cryptococcal antigenemia). This condition may be more common than symptomatic disease, and patients may develop clinical cryptococcal disease unless treated and so are now included in these definitions. The term “disseminated cryptococcosis” as distinct from CNS cryptococcosis has been abandoned in favor of the terms “pulmonary,” “CNS” and “other extrapulmonary sites.” “Colonization” with Cryptococcus spp. is no longer included in the definitions as it is poorly understood and its natural history is unknown.

19Identification to the species level for Cryptococcus neoformans and Cryptococcus gattii has become increasingly important based on reports that suggest different clinical presentations, outcomes, and responses to antifungal therapy between these 2 species. Verification of species that use CGB (L-canavanine, glycine, bromthymol blue) agar or matrix-assisted laser desorption ionization–time of flight mass spectrometry is recommended. Outcomes for HIV-associated cryptococcosis due to C. neoformans and C. gattii appear to be similar, and identification to the species level may be unnecessary.

20The endemic mycoses are caused by environmental fungi that are usually restricted geographically and cause disease in immunocompetent and immunocompromised hosts. Causative agents include Histoplasma capsulatum var. capsulatum and H. capsulatum var. duboisii, Blastomyces species complex (eg, B. dermatitidis, B. gilchristii, B. helicus, B. silverae, and B. parvus), Coccidioides immitis/Coccidioides posadasii, Paracoccidioides brasiliensis/Paracoccidioides lutzii, Sporothrix species complex (S. brasiliensis, S. schenckii sensu stricto, S. globosa, and S. luriei), Talaromyces (formerly Penicillium) marneffei, and Emergomyces species (E. pasteurianus, E. africanus, E. orientalis, E. canadensis, and E. europaeus). Probable endemic mycoses are defined by evidence of environmental exposure to the fungus, a compatible clinical illness, and the presence of either Histoplasma or Blastomyces antigen in any body fluid or antibody to Coccidioides species in serum or CSF as recovery by culture and histopathologic evidence of infection is generally lacking. There are no approved serologic tests for T. marneffei, S. schenckii species complex, or P. brasiliensis. Exposure to 1 of these fungi is defined as a history of residence in an endemic area, no matter how remote, travel to an endemic area, or contact with fomites such as soil or vegetation that is derived from an endemic area.

21The revised definitions of proven IFD are shown in Table 1.

22Several changes were made to the definitions of probable IFD (Tables 2 and 3). For example, host factors were expanded to include inherited severe immunodeficiency and low CD4 lymphocyte counts. Radiographic features were expanded to include wedge-shaped and segmental or lobar consolidation and a reverse halo sign to indicate mold disease of the lower respiratory tract. Revised thresholds for GM now replace those of the manufacturer. Aspergillus PCR is now included, and there are mycologic criteria for non-HIV–associated pneumocystosis.

23While definitions of proven and probable disease have been shown to be reliable in research and attracted little controversy among the group, this cannot be said of the possible IFD category. There is much confusion about the difference between a host factor and a risk factor. As before, a host factor has been defined as a characteristic of individuals clearly predisposed to, and not simply at risk of, an IFD. For example, while impaired gut wall integrity through surgery or illness may increase the risk of candidiasis, it was not considered specific enough to warrant inclusion as a host factor. Pulmonary abnormalities such as tree-in-bud opacities and interstitial abnormalities were excluded from the clinical features as they can be due to a wide range of pathologies in addition to IFD.

24Throughout this process, we have emphasized the need to differentiate between definitions of IFD required for clinical research from those that influence clinical practice. In clinical practice, many would administer an antifungal agent to any patient at risk of IFD when fungi are detected by biomarkers in serum, plasma, whole blood, or relevant body site fluid without there being sufficient evidence to satisfy the consensus definitions of IFD. We also recognize that our definition of a host factor errs on the side of conservatism given the increasing use of drugs such as monoclonal antibodies for treating a variety of conditions.

25Other controversial issues included distinguishing between the performance characteristics of tests for screening and confirmation, the impact of exposure to antifungal agents used for prophylaxis or treatment on imaging and diagnostic tests, and the use of biomarkers to monitor therapeutic outcome. We agree that further research will be required to evaluate the evidence for each of these assays. Finally, there was consensus that diagnostic strategies to determine the relative efficiency of an available test, alone or in combination with other diagnostic tests, should be considered further.

26In summary, these revised definitions represent consensus expert opinion based on the best available evidence. As such, they will need to be reviewed regularly for their utility and relevance and, where possible, extended to other populations affected by IFDs. We acknowledge the limitations of these definitions, including the exclusion of certain cases of IFD. However, the reliance on host factors, clinical features, and mycologic evidence to define IFD in selected populations has proven its value for clinical trials, epidemiologic studies, and the evaluation of diagnostic tests.