Towards molecular diagnostics of invasive fungal infections

Invasive fungal infections (IFI) are an important cause of severe illness in immunocompromised patients, particularly individuals with malignant disorders undergoing high-dose chemotherapy, allogeneic stem cell transplant recipients, preterm neonates, intensive-care patients and individuals with acquired or innate immune deficiencies. Since the clinical symptoms of IFI are often nonspecific [1], fungal pathogens are not identified as the causative agents in a notable proportion of cases or the diagnosis is established late in the course of infection. This contributes to high mortality rates, reaching 40–60% [2,3] and even exceeding 90% in the presence of CNS involvement [4]. Most cases of IFI are still attributable to Candida and Aspergillus species [4,5] but the epidemiology of fungal infections has been changing over the last few years [1,5–7]. This is, at least in part, attributable to the broad application of antifungal prophylaxis in high-risk patients, which has resulted in the increased occurrence of resistant species or the emergence of hitherto uncommon fungal pathogens displaying inherent resistance to antifungal agents used in the prophylactic setting. Examples of this phenomenon include:

• • Increased infection rates by non-albicans Candida species, such as Candida krusei and Candida glabrata, which are resistant to fluconazol or by Aspergillus terreus which is resistant to amphothericin B [6,8];

• • More frequent occurrence of zygomycoses as a result of prophylaxis by voriconazol [1,6,9], which is active against fluconazol-resistant Candida species and amphothericin-resistant A. terreus, but ineffective against Zygomycetes.

Hence, changes in the spectrum of fungal species involved as emerging pathogens in IFI greatly increase the demands on diagnostic tools used in the clinical monitoring of immunocompromised patients. In this regard, the currently available options for the detection and surveillance of fungal infections are not satisfactory. Owing to the limitations of current diagnostic approaches, prophylactic and empirical treatment concepts still represent the standard of care in severely immunocompromised patients.

Microbiological cultivation has limited sensitivity and specificity, and may often take too long to be clinically useful [10,11]. Blood cultures were described to fail in a significant proportion of candidiases [12,13]. Moreover, the detection of aspergillosis from blood cultures is rarely successful, and positive results cannot be readily associated with invasive aspergillosis [14]. Serological tests detecting fungal cell wall components, such as galactomannan or (1,3)β-D-glucan have become widely used diagnostic tools. The galactomannan sandwich ELISA, which permits the detection of Aspergillus (and reportedly also Penicillium[10]) antigens, was described by different authors to offer good specificity but variable sensitivity [15,16]. The occurrence of false-positive results of galactomannan-based detection tests was described in patients receiving antibiotic treatment with semisynthetic penicillins [17,18] or contaminated plasma preparations [19,20], and particularly in children, apparently in relation to the diet [21]. In contrast to galactomannan, (1,3)β-D-glucan is present in the cell wall of most fungi, with the exception of Cryptococcus spp. and Zygomycetes, thereby providing a broader detection spectrum of fungal pathogens [10]. False-positive results of (1,3)β-D-glucan-based detection assays were described in patients treated with immunoglobulin or albumin preparations contaminated with fungal components, in recipients of products filtered through cellulose filters [10,22] or in patients suffering from bacterial infections (e.g., certain streptococci are known to produce glucan or glucan-like polymers) [23]. Tests relying on (1,3)β-D-glucan detection appear to be more sensitive than galactomannan detection assays in patients with invasive aspergillosis [22,24] but none of the presently available serological tests provides information on the fungal species present. The intrinsic lack of mycological specificity requires the integration of clinical, radiological and microbiological data for proper interpretation [22]. Despite the apparent shortcomings, the overall assessment of serological testing by a consensus group of the European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group (EORTC) and Mycosis Study Group of the National Institute of Allergy and Infectious Diseases (MSG) has led to the inclusion of seropositivity in the clinical diagnostic criteria for IFI [12,25]. Imaging methods play an important role in the diagnosis of IFI but radiological findings are often nonspecific, and may not be conclusive until late in the course of disease [15]. CT scanning can detect invasive mycoses, such as aspergillosis, at an earlier stage of infection. However, the typical signs were described to be less sensitive in non-neutropenic, corticosteroid-treated patients than in patients displaying neutropenia. This observation possibly reflects the different inflammatory response with decreased fungal load and increased inflammatory cellular trafficking in the non-neutropenic lung [15]. Histopathological examination of targeted biopsies of suspicious lesions identified by imaging methods is an important parameter for the diagnosis of proven fungal infection. However, the presence of profound thrombocytopenia in immunocompromised patients often prevents the performance of biopsies owing to the fear of bleeding complications.

The current diagnostic options, therefore, often fail to provide a rational basis for antifungal therapy, and there is urgent need for improvement. The introduction of molecular methods for the detection of fungal infections was an important step in this direction. A major advantage of these methods is the high sensitivity and rapid availability of results, which could serve as an important prerequisite for timely onset of antifungal therapy. At present, different molecular platforms are available for qualitative and quantitative fungus detection and for species identification. Examples of commonly used formats include standard PCR [26–28], PCR-ELISA [29–31], different variations of real-time PCR [32–37], nested PCR [38–40], multiplex PCR followed by DNA microarray [41,42], nucleic acid sequence-based amplification [43], fragment size analysis of variable regions in the fungal genome [44], DNA sequencing [28,45], hybridization to specific capture probes bound to microbeads (Luminex technology) [46–48] and pyroseqencing [49,50]. Currently, PCR-based assays are the most commonly used technical approach to molecular fungus analysis permitting either species-specific, genus-specific or broad-range fungus detection and, depending on the specific format of the assay, identification of the fungal species and quantification of fungal load. The most commonly targeted genes of molecular fungus detection assays include the ribosomal genes (18s rDNA, 28s rDNA, ITS2 and ITS1 regions) and, less frequently, mitochondrial genes. The preferred target genes are present in multiple copy numbers in the fungal genome, thereby increasing the sensitivity of detection. Most of the current molecular methods are focused on the detection of Aspergillus and Candida spp., and the number of assays that cover a broader spectrum of fungal pathogens, also including the ‘emerging fungi’ (e.g., Zygomycetes and Fusarium spp.), whose occurrence in IFI has been increasing, is rather limited [28,36,42,48,51]. Although molecular assays undoubtedly have great potential for improvement of IFI diagnostics, a number of variables affecting their clinical applicability require careful attention before these tests can be deemed eligible as a basis for treatment decisions. The issues of concern include the choice of appropriate clinical materials, the required sample size, the efficacy and safety of fungal nucleic acid isolation, the availability of standardized methodologies for molecular fungus detection, the required test performance in terms of sensitivity, specificity and other relevant parameters, the validation of the techniques in clinical studies, and the correct interpretation of test results.

The types of clinical specimens most commonly used for molecular testing in patients at high risk of IFI include serum, plasma and whole blood, but also bronchoalveolar lavage, and fresh or paraffin-embedded tissue from affected sites. For PCR assessment of invasive aspergillosis, whole blood was claimed to be superior to plasma [52], but for the detection of invasive candidiasis, serum specimens were described to be preferable to whole blood [53]. This example illustrates the current lack of firm knowledge regarding the appropriate selection of clinical materials providing the most suitable source for timely molecular diagnosis of IFI. Clinical studies addressing this issue are clearly needed. The range of sample sizes used for molecular analysis is very broad (200 µl to 10 ml [54]) and, currently, no recommendations on adequate sample volumes are available. In order to enable addressing this issue, it is necessary not only to determine the efficacy of individual DNA isolation and detection protocols, but also to assess the clinical requirements for the sensitivity of molecular fungus detection. It is well known that the concentration of fungal particles in the peripheral blood of patients with IFI is usually very low [32]. Protocols for the isolation of fungal DNA that permit processing of larger volumes of starting material (in the range of milliliters) can, therefore, be expected to be advantageous for clinical testing. Most of the commonly used protocols are based on enzymatic and/or mechanical pretreatment facilitating efficient disruption of the fungal cell wall followed by different methods for DNA purification including phenol-chloroform, affinity spin-columns or (semi-)automated magnetic bead extraction [54]. A recent multicenter study showed that satisfactory results can be obtained by a variety of technical approaches, but the use of spin-columns is more prone to carry-over contamination [55]. In view of the fact that contamination by airborne spores or extraneous fungal DNA represents a serious problem in molecular testing of pathogenic fungi, a number of measures are required. Sterile collection of clinical specimens and sample processing in laminar air-flow units are important prerequisites for reliable test results, and the use of closed formats for DNA extraction (e.g., the MagnaPure system) may further reduce the risk of contamination. The observation that a variety of materials and commercially available reagents may contain traces of fungal DNA requires careful testing of all new solutions and reagent batches before they can be used for clinical analysis. For a number of specific applications, appropriate reagent grades can be obtained, and it would be highly desirable to have commercially available ‘fungal-grade’ materials and reagents, which would be completely free of any fungal contaminants, in order to increase the safety and reliability of molecular testing of pathogenic fungi.

A plethora of methods for the molecular detection of IFI, including a number of assays with broad specificity, have been published to date, but their standardization across a wide spectrum of diagnostic laboratories is lacking. Owing to the apparent absence of adequately standardized techniques, molecular methods are currently not accepted by the EORTC as a clinical diagnostic parameter for the assessment of IFI, although the great potential of this approach is acknowledged [25]. Multicenter efforts directed toward the standardization of molecular fungus detection assays are underway [55]. The international harmonization of diagnostic approaches is an important prerequisite for large-scale multicenter studies addressing the adequacy of molecular screening and monitoring for clinical diagnostics of IFI. A major difficulty inherent in this task is the lack of a gold standard for the diagnosis of fungal infections in relation to which molecular assays could be evaluated. The diagnostic criteria for proven, probable and possible IFI established by the EORTC are widely accepted but, in many instances, the available test results only permit the diagnosis of possible or probable infection, thus rendering the evaluation and clinical validation of molecular diagnostic assays difficult. Recently, a meta-analysis of PCR methods for the diagnosis of invasive aspergillosis has been performed, and might be the first step towards the aim of clinical validation [54].

Our own results based on serial screening of peripheral blood specimens in high-risk pediatric patients by a panfungal real-time PCR assay established in our laboratory [101] indicate that molecular testing has a high sensitivity for the detection of IFI [51], which has also been suggested by other groups [32,37]. However, the specificity of the assay is difficult to interpret owing to the low proportion of patients identified as having a proven IFI according to the EORTC criteria. As already observed by others, a single PCR-positive test is of questionable value, but two subsequent positive results may provide relevant information [54]. Nonetheless, in order to establish a firm basis for clinical validation of molecular analyses, there is urgent need for clinical studies combining molecular testing with established diagnostic methods permitting the identification of proven fungal infections, such as histopathological examination of targeted biopsies and post-mortem analysis of organ involvement. The validation of standardized molecular assays in carefully designed clinical studies would be an important prerequisite for improvement of the diagnostic armamentarium, which is required for the establishment of rational approaches to effective pre-emptive antifungal therapy. The implementation of molecular diagnostic strategies could permit timely onset of appropriate treatment, and may be expected to pave the way for improved clinical outcome of invasive fungal infections.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.