A longstanding diagnostic challenge for fungal infections is to identify an infecting fungal pathogen early in the infection process and rapidly assess its susceptibility to antifungal drugs to facilitate appropriate therapy. Conventional culture-based approaches are too slow, often requiring several days, and insensitive which has led to the development of modern molecular approaches. The molecular revolution in biology has allowed some impressive developments in mycology diagnostics. The principal uses of molecular methods in the diagnostic mycology lab include:
- Direct detection of fungal pathogen nucleic acid for diagnosis
- Identification of positive cultures to genus or species level
- Molecular typing and phylogeny for outbreak and cluster analysis
- Rapidly identify the mechanisms of antifungal resistance
The last 2 methodologies require highly specialised labs, and so are not covered here, although valuable and important to understand disease transmission and selecting appropriate therapy.
Polymerase chain reaction (PCR) and real-time detection is an important molecular method for diagnosis that amplifies and accurately identifies fungal-specific nucleic acid for diagnosis, but sample type, sample preparation and carefully controlled laboratory conditions are essential for optimal results and to prevent both false positive and negative results. Test performance and cut-off values vary between different patient types and samples. Some tests are commercially available and partially or completely validated. Some are offered as a service and have usually been analytically validated but not always clinically validated. Clinical validation is often challenging and time consuming as each patient group and sample type needs testing, and the current methods for confirming the diagnosis lack precision.
Some drugs and other substances can inhibit PCR reactions and so all diagnostic assays should include an amplification control to exclude a failed (false negative) reaction due to inhibition. This is standard practice for commercial assays but not for some in house assays. Negative control assays are also important to rule out both environmental and amplicon (ie previous PCR assay) contamination, giving rise to false positive results.
Direct detection of dermatophyte and cutaneous fungal DNA
Microscopy and culture of nail, skin and hair samples is the usual labour intensive and slow means of establishing the diagnosis of a dermatophyte infection. Molecular diagnosis using kits for dermatophytes and Trichophyton rubrum generate an answer faster and are 4-18% more sensitive than conventional diagnosis. Time from sample to receipt and result is can be as little as 5 hours, but is generally 24 hours. One conventional CE marked PCR kit developed at the Serum Statens Institut detects all dermatophytes and T. rubrum, the most common cause of skin and nail infections. SSI Diagnostica. MycoDerm, another CE marked kit sold by Biotype Diagnostic GmbH detects 21 dermatophytes, yeasts and moulds from clinical specimens Biotype. Both kits include simple nail dissolution methods. A third kit CE marked real-time PCR kit from Fast-Track Diagnostics detects and identifies 4 Trichophyton species and 3 Microsporum species directly Fast Track Diagnostics.
Direct detection of Aspergillus spp. nucleic acid for diagnosis
While Aspergillus spp. can sometimes be cultured from respiratory secretions in patients with invasive, chronic and allergic aspergillosis, the yield is poor and from blood it is rare with a frequency of <1%. Molecular diagnosis offers a more sensitive and potentially faster means of detectingAspergillus. Subsequent sequencing of positive samples may provide species information, which is lacking with other biomarkers such as galactomannan and beta 1,3-glucan. The main challenge with achieving sufficient sensitivity to detect Aspergillus (and other filamentous moulds) is getting a high quality sample, and efficient DNA extraction system and a highly sensitive real-time PCR reaction. Preventing false positive results because of contamination in the sample container or extraction reagent Aspergillus environmental DNA contamination is the major difficulty with getting a fully operational and useful system in the clinical microbiology laboratory.
The most common regions for detection are the 18S rRNA, 28S rRNA and ITS regions. All are multicopy genes with 35-90 copies per nuclear genome. These targets provide natural amplification, improving sensitivity of detection. However some primer and probe selections exclude some common Aspergillus species, or only include A. fumigatus. Many systems cannot distinguish the very closely related Penicillium spp. slightly impairing specificity. Commercially available assays (2012) include MycAssay Aspergillus (Myconostica), Septifast (Roche), VetPCR ASP.FUM Detection (Veterinary) (BioinGentech), MycoReal Aspergillus (Ingenetix GmbH), Affigene Aspergillus tracer (Cepheid) and Aspergillus spp. Q-PCR Alert (Nanogen), RenDx multiplex Aspergillus spp & Candidasp(whole blood, plasma & serum). In a modest direct comparison, MycAssay Aspergillus was superior to Aspergillus spp. Q-PCR Alert. Other comparisons are lacking. Some kits are designed only for blood, others for respiratory or other samples, primarily because of differences in sample fungal DNA extraction, necessary for clinical validation.
On blood (and there are many different ways of extracting DNA from whole blood, clot, serum and plasma) in haematology patients, meta-analyses or real-time PCR suggest a sensitivity of ~75% and negative predictive value (NPV) over 95% for 2 positive or negative samples. Sensitivity is much lower in non-haematology patients. These performance characteristics compare well with galactomannan. An area of uncertain is the relative performance with and without antifungal prophylaxis; some infections can be diagnosed despite itraconazole prophylaxis and empiric therapy with PCR. Furthermore, the application of PCR following therapy to assess disease burden is uncertain.
For respiratory samples, PCR is more sensitive than culture in the context of multiple disease types. Sample source matters and very dilute bronchoalveloar lavage (BAL) samples have lower fungal loads than upper airway and sputum specimens. Cystic fibrosis samples require liquefaction to optimize yield since respiratory samples are highly viscous and have a complex mix of host and bacterial nucleic acid.
Tissue samples can also be analysed by real-time PCR. Fresh tissue is easier to extract than fixed tissue but both yield satisfactory nucleic acid for analysis. Probably other samples are also useful such as corneal scrapings and vitreous, but there is limited published experience. Given the poor sensitivity of culture, it is likely that PCR will become standard practice for these special samples.
Direct detection of all fungal DNA (pan-fungal) for diagnosis
In certain clinical situations, the likelihood of a specific fungal infection is high and a directed molecular test is useful, as for example detecting P. jirovecii. However in numerous other situations, there is a wide differential diagnosis with regard to which fungus, if any, is responsible. For this reason there have been several reports of using panfungal detection, attempting to pick up all possible fungi responsible. As the DNA databases have expanded with additional sequences being deposited in GeneBank and other repositories, so the precise specificity of primers and probes can be better interrogated. The sensitivity of any panfungal assay is likely to vary somewhat according to the fungus being detected (with slightly different reaction kinetics for different fungi), as well as extraction and internal copy number differences between different genera, species and strains. Overall a panfungal assay could have utility if it has a high negative predictive value to exclude a fungal infection, or as a ‘capture’ assay so that the precise fungus responsible can be identified by second step sequencing. One very specific problem with panfungal assays is the contamination issues for all reagents in the assay and all plasticware. False positives are problematic in frequency.
Only one commercial panfungal assay is available (2012), namely MycoReal Fungi (Panfungal PCR;Ingenetix GmbH, Vienna). Nothing is published about this assay.
Identification of cultures or tissue for identification
Several molecular identification kits are available (2012) for identifying cultures. These include MicroSeq D2 LSU rDNA Fungal Identification Kit (Applied Biosystems), BlackLight® Fungal ID Kit (BlackBio), Pyrosequencing (Qiagen), AccuProbe kits for the identification of Blastomyces dermatitidis, Histoplasma capsulatum and Coccidioides immitis. PNA FISH methodology (AdvanDx) provides a partial indication of Candida species in blood culture.
Sometimes a fungus is seen in a specimen or tissue and not cultured. It is important to identify such fungi to genus and preferably species level. Fresh non-embedded tissues have shown that sensitivity for PCR detection of fungi exceeds 95%, while the sensitivity of paraffin-embedded samples is currently ~60%. If fungal infection is strongly suspected prior to biopsy or resection, retention of some of the sample fresh (ie not placed in formalin) may facilitate aetiological diagnosis. The fungal DNA extracted from FFPE specimens can be degraded and in low concentration, and it often contains substances that inhibit protein digestion or DNA amplification. However, when fungal elements are detected in FFPE tissue sections and fungus culture is not available, PCR can in some cases determine the organism that is causing the infection.
Only for Aspergillus spp. is a commercially available technique to determine the genus of fungi found in tissue sections published, but this method does not separate species of Aspergillus. The majority of the published assays target specific rRNA genes (18S or D1-2 of 28S) or the intervening internal transcribed spacer (ITS1 and ITS2).
A common strategy is to amplify up one or two discriminatory regions (such as ITS1 and D1-2 of 28S), and sequence these. For some species, other single copy genes appear to more discriminatory for determining species, such as calmodulin, but this approach has rarely been applied to tissue or microscopy positive samples.
One of the major challenges currently is the relatively low quality of the databases for bioinformatic comparison. Both, sequence errors and speciation mistakes conspire to make results less than reliable. For this reason several academic groups and reference laboratories have built up their own databases to improve this situation. These are not currently available online, although this situation is being addressed.