Recognition of Diagnostic Gaps for Laboratory Diagnosis of Fungal Diseases: Expert Opinion from the Fungal Diagnostics Laboratories Consortium (FDLC)

Disease-specific gaps. (i) Pneumocystis pneumonia: develop an optimal diagnostic algorithm.

Pneumocystis jirovecii is a yeast-like fungus that causes Pneumocystis pneumonia (PCP), an opportunistic lung infection associated with high mortality in immunocompromised patients (1, 2). National and international societies have recently published guidelines which include the diagnosis of PCP in non-HIV immunocompromised patients (3, 4). For decades, the gold standard for diagnosing PCP has been direct microscopic identification (ID) of P. jirovecii cysts and trophozoites in tissue samples, induced sputum, and bronchoalveolar lavage (BAL) fluid using nonspecific stains such as Grocott-Gomori methenamine-silver, calcofluor white, or a specific stain with fluorescence monoclonal antibodies against P. jirovecii. Direct visualization techniques of P. jirovecii cyst and/or trophic forms have inferior diagnostic sensitivity in non-HIV patients, which can lead to false-negative test results and missed diagnoses (5).

In recent years, NAATs of respiratory samples have enabled direct detection of P. jirovecii DNA (6–8). A P. jirovecii NAAT is appealing due to higher sensitivity than direct microscopic methods, especially in non-HIV-infected patients. Different P. jirovecii NAAT chemistries and gene targets have been described, but most of the P. jirovecii NAAT assays were developed in-house as laboratory-developed tests (LDTs), and it is hard to compare the diagnostic performance of each assay without a standardized approach (9, 10). Furthermore, NAATs may be prone to misinterpretation in cases where P. jirovecii is a colonizer (i.e., present without causing disease) rather than the cause of disease (i.e., present in the context of clinical signs or symptoms of acute pneumonia) (11–14). The use of a quantitative P. jirovecii NAAT (qNAAT) has been proposed to differentiate between colonization and disease state (3, 4). The cycle threshold (C_T) values obtained by using a qNAAT allow estimation of the fungal burden with the assumption that a higher fungal burden increases the probability of a disease state. However, caution needs to be taken when applying C_T values to correlate the organism load with the disease severity because different factors, including test method variation, may influence C_T values (15). Detection and quantification of P. jirovecii DNA using qNAAT in conjunction with establishment of population-specific C_T thresholds might enable differentiation between colonization and infection (16–18). However, this approach is limited by a lack of standardization across P. jirovecii qNAAT tests, which impedes the establishment of universal interpretive thresholds.

1,3-β-d-Glucan (BDG) is an antigenic component of the cell wall of many fungi, including P. jirovecii, and is considered a diagnostic adjunct for PCP (19, 20). Although it lacks specificity due to its elevation in various fungal infections, serum BDG testing is often used as a noninvasive means to support the diagnosis of PCP, especially in situations where critical illness precludes invasive diagnostic procedures (20–22). However, the optimal test cutoff value for the diagnosis of PCP still remains to be determined (23). In addition, more data are needed to evaluate the negative predictive value of BDG testing.

According to the European Organization for Research and Treatment of Cancer and the Mycoses Study Group (EORTC/MSGERC) guidelines, the criteria for “proven” PCP remains the detection of cysts in tissues, BAL fluid, or sputum using direct microscopy, while “probable” PCP now includes the use of BDG and NAAT (3). Given the limitations of both BDG and NAAT as described above, an approach that combines both methods could increase the accuracy of PCP diagnosis. In this option, both BDG and P. jirovecii NAAT are performed, and the combination of results were interpreted depending on the clinical context and the patient population (24, 25). The development of the best gene target and standardized commercial P. jirovecii qNAAT assays and/or quantitative international calibration standards would enable meaningful evaluation of C_T thresholds for defining colonization versus disease state. In addition, future multicenter studies are necessary to bridge the current gap in knowledge surrounding the optimal role of molecular detection of P. jirovecii and serum BDG testing either alone or in combination with diagnostic algorithms for PCP in distinct patient populations.

(ii) Mucormycosis (formerly known as zygomycosis).

(a) Reevaluation of optimal tissue processing for isolation of Mucorales. Processing tissue specimens submitted for fungal culture can be done by mincing (cutting a tissue into small pieces) or homogenizing (grinding the whole tissue into suspension) before inoculating onto culture plates. Major clinical microbiology reference texts give particular emphasis to the importance of mincing instead of homogenizing tissue when infection with a member of the mucoraceous molds is suspected (26–28). This is purportedly to avoid damage to fragile hyphal structures of the Mucorales molds, although there is no published data to support this notion. No studies are cited to support this practice, nor could we find such studies when we searched the scientific literature. Indeed, the anecdotal experiences of several laboratories call into question this practice based on the observation that nearly 50% of the tissues that were positive for Mucorales hyphae in calcofluor white stain or in histology yielded no growth in culture using the mincing method for tissue processing (unpublished data). Occasional observations have been made in cases where the tissue submitted for bacterial culture (which was homogenized in a tissue grinder) grew a Mucorales mold, while the tissue concurrently submitted for fungal culture (which was minced) did not yield any fungal growth.

Formal studies are lacking that directly compare the effect of tissue mincing versus grinding on recovery of Mucorales. Homogenization of tissue usually results in releasing microorganisms into the suspension suitable for downstream culture. This tissue process does not show any damage to bacterial culture recovery and actually is recommended for culture isolation of Histoplasma; its detrimental effect on the Mucorales molds is therefore called into question by anecdotal experiences. These observations warrant further multicenter studies to directly compare mincing versus grinding in order to discern the best tissue processing strategies to enhance the diagnostic yield of fungal culture for Mucorales.

(b) Develop Mucorales NAAT for early diagnosis of mucormycosis. Early diagnosis of mucormycosis is key to increase survival outcomes. There are no commercially available Mucorales antigen tests in serum but large amount of circulating cell-free Mucorales DNA in blood has been detected as early as up to 9 days prior to the diagnosis made by culture in patients with mucormycosis (29–32). Therefore, NAAT-based tests may be the most sensitive method to detect Mucorales DNA in blood and serve as a screening and diagnostic assay for early diagnosis. NAATs have been developed for both BAL and blood samples. Depending on the test, results could be available in <2 h (33, 34). Early diagnosis of pulmonary mucormycosis in patients with hematological malignancy has been achieved through detection of Mucorales DNA in both serum and whole-blood samples using a pan-Mucorales NAAT assay targeting 18S ribosomal DNA (rDNA) (30, 31). Another Mucorales-specific gene target is the CotH gene encoding spore coat homolog proteins. In both animal model and human infection cases, an NAAT assay targeting the CotH gene demonstrated early diagnosis of mucormycosis with better sensitivity seen in urine samples than in BAL and blood samples (35). Despite all efforts made so far to develop sensitive molecular assays to aid in early diagnosis of mucormycosis, Mucorales NAAT is not included in the recent revised EORTC/MSG criteria for probable invasive mold infection (3). In a recent global guideline for diagnosis and management of mucormycosis, detection of Mucorales DNA in serum, as well as in other body fluids, is supported only with moderate strength (36). Lack of standardization and relative rarity in cases prevent implementing a molecular assay in clinical laboratories to improve the speed and sensitivity for diagnosing mucormycosis. Variable factors such as sample type, method of extraction, NAAT target, and NAAT format and conditions contribute to the performance of the NAAT assay. Development of a commercial assay could potentially standardize these variables and facilitate laboratory validation and implementation. Currently, there is only one commercially available Mucorales NAAT assay (MucorGenius; PathNostics) that demonstrated early diagnosis of mucormycosis (32, 34), but the assay has not been approved by the FDA. Multicenter investigations seeking to standardize Mucorales NAAT and assess its clinical utility may improve diagnosis of this rapidly progressive and often fatal infection (37).

(iii) Aspergillosis: NAAT in conjunction with galactomannan for early and accurate diagnosis.

Microbiologic diagnosis for invasive aspergillosis (IA) relies on recovery of Aspergillus spp. in culture, microscopic examination of sterile samples, and detection of the galactomannan (GM) antigen in serum and BAL fluid. Although useful, these methods lack specificity, are time-consuming, and can result in inconclusive findings. The GM antigen assay is most useful in neutropenic leukemia patients but lacks sensitivity in high-risk hematopoietic stem cell transplant patients on antifungal prophylaxis or in other nonneutropenic immunocompromised patients (38, 39).

The diagnostic utility of NAAT for the diagnosis of IA remains unknown. Although NAATs targeting Aspergillus spp. from BAL and blood specimens are commercially available (40), the majority of NAATs for Aspergillus species are still LDTs (41). Most utilize real-time PCR (RT-PCR) to detect Aspergillus at genus level, but some additionally identify individual species yielding results for: A. fumigatus, A. terreus (amphotericin B resistance), and other Aspergillus spp. (42, 43). Aspergillus NAAT is widely used in Europe and recommended in ESCMID guidelines (44, 45). The most recent EORTC/MSG criteria includes Aspergillus NAAT in defining proven (NAAT on tissues) or probable IA (NAAT on blood and BAL samples) (3, 38).

The potential advantages of an Aspergillus NAAT include a rapid turnaround time, increased clinical specificity compared to GM, the ability to differentiate between Aspergillus species, and detection of antifungal resistance markers (46). Disadvantages include the need for specialized equipment, molecular expertise, and false-positive results caused by ubiquitous Aspergillus spp. transiently colonizing the respiratory tract or contaminating reagents (e.g., silica in nucleic acid extraction columns; sodium citrate vacutainers). Like culture, detection of Aspergillus DNA from nonsterile sites does not necessarily equate to invasive infection. Quantitative NAAT applied to BAL fluid has been explored as a means to increase the predictive value of the NAAT. While higher fungal burdens may be more suggestive of IA, there remains significant overlap in DNA loads between colonization and invasive disease (47).

Given substantial interlaboratory NAAT performance variability, the European Aspergillus PCR Initiative (EAPCRI) has led efforts to optimize and standardize Aspergillus NAAT testing (48–52). Commercial NAAT assay may also provide standardization and accessibility; AsperGenius (PathoNostics) is the only one commercially available that has been studied and reported with good clinical performance (42). Screening for Aspergillus infection in blood samples from high-risk populations not receiving prophylaxis can be done using NAAT (84 to 88% sensitivity; 75 to 76% specificity) or GM (79 to 80% sensitivity; 81 to 86% specificity) (53). An NAAT exhibits higher sensitivity and identifies more patients at risk of IA requiring additional testing (physical exam, chest computed tomography, and BAL analyses) (53). Screening patient blood using NAAT and GM increases the sensitivity to 99% (when at least one test is positive) and specificity to 98% (when both tests are positive), leading to decreased use of empirical antifungals, earlier detection, and reduced mortality (53–56).

Multicenter clinical trials using optimized testing protocols are needed to evaluate the utility of the Aspergillus NAAT in conjunction with GM. Twice-weekly screening with serum NAAT and GM in high-risk populations with hematologic malignancy is expected to mitigate prophylactic antifungal use while enabling prompt treatment initiation with high sensitivity. Likewise, the combined use of GM and NAAT on BAL samples from symptomatic individuals may increase diagnostic accuracy. As part of well-designed prospective studies, additional work is required to determine the clinically relevant quantitative thresholds that best differentiate patients with IA and low levels of fungal DNA from specimens that may be contaminated or have high background signal.

Azole resistance in A. fumigatus is increasing, attributable to mutations (L98H, TR34, T289A, and Y121F) in the CYP51A, the gene encoding lanosterol 14α demethylase (57, 58). Multicenter evaluations on the clinical utility of NAAT to rapidly detect A. fumigatus CYP51A azole resistance mutations and species, such as A. terreus, with clinically actionable innate resistance patterns are needed and would be expected to show more rapid time to appropriate therapy relative to traditional methods given low culture sensitivity and long ID and mold antifungal susceptibility testing (AST) turnaround times. Opportunities to incorporate the detection of cryptic Aspergillus species, such as A. lentulus or A. calidoustus, with reduced azole susceptibility may also be of significant value.

(iv) Candidemia: optimize rapid test algorithm.

Candidemia is the third to the fourth most common bloodstream infection in health care settings, and the predominant severe fungal infection developing in critically ill patients in intensive care units (ICUs) (59). Other patients with malignancy, transplantation, immunosuppression, abdominal surgery, prolonged broad antibacterial use, and injection drug use are at high risk of developing candidemia (60–62). Overall crude mortality rate during the hospitalization with candidemia is approximately 25% (63). Delayed diagnosis and initiation of inappropriate antifungal treatment are associated with higher mortality rates (64).

Recent progress has been made to speed up the detection and ID of Candida spp. that cause candidemia. Details on the FDA-cleared methods for the rapid ID of Candida spp. from blood are summarized in Table 1. The most widely adopted methods involving nucleic acid detection assays applied to aliquots from positive blood culture bottles, showing highly accurate and significantly speedy time to organism identification (65–68). Matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry (MS) has also been used directly on positive blood culture aliquots (69, 70). This approach requires blood lysis and protein extraction up front, which adds additional hands-on time and complexity. In theory, MALDI-TOF MS incorporating robust spectral databases would be expected to be able to identify the broadest spectrum of Candida species compared to the targeted molecular methods.

Current commercial blood culture systems are estimated to detect invasive candidiasis (IC) with a sensitivity of approximately 50% (71). While the aforementioned commercially available rapid diagnostic tests are capable of expediting identification of the Candida spp., they still rely on blood culture turning positive, and thus they may not increase the speed and sensitivity for the detection of IC. A single platform (T2Candida) has been FDA cleared for the direct detection of the five most common Candida spp. in whole blood, greatly speeding the time to candidiasis diagnosis (72, 73). However, given the limited number of organisms detected by T2 combined with a clinical sensitivity of approximately 90%, routine blood culture is still required as a back-up. Direct detection of Candida species from blood using molecular methods may increase sensitivity compared to culture but its clinical benefits and costs are unclear (74). Developing a sensitive molecular test of blood also needs to take into consideration the prevalence of candidemia in a patient population. For example, in settings with low prevalence of candidemia in a general inpatient population (e.g., ≤1%), PPV of T2Candida has been estimated to range from 15 to 31%, which does not strongly support its use (74). Multicenter controlled clinical trials enrolling the optimal patient population for direct detection of Candida would help to maximize the clinical outcome and impact (e.g., reductions in length of hospital stay and mortality rate) of direct testing and contain costs. In addition, coupling rapid Candida identification with rapid phenotypic antifungal susceptibility testing may help to optimize antifungal use.

(v) Endemic mycoses: improve availability and performance of diagnostics.

The endemic mycoses in North America are histoplasmosis, blastomycosis, and coccidioidomycosis. Since each of the dimorphic fungi has a morphologic mimic when grown in culture, microscopic preparations with features suggestive of an endemic fungus should be confirmed with a genetic probe (i.e., AccuProbe; Hologic, Inc., San Diego, CA [this commercial product may be phased out soon]) or another molecular method (e.g., DNA sequencing). It has been demonstrated that MALDI-TOF MS was able to identify Histoplasma, Coccidioides from the mycelial phase of the colonies (75, 76), but biosafety precaution needs to be taken due to potential laboratory exposure to the organisms.

Genus- and/or species-specific NAAT assays are available that target endemic fungi (77–90). Although useful to confirm the presence of a pathogen in respiratory specimens and tissues, these assays have not been widely adopted. In contrast, a broad-range NAAT with 28S rRNA sequencing, which enables the identification of most fungi in clinical specimens, exhibits a higher diagnostic yield (91–94). Importantly, these methods are not a replacement for culture, as demonstrated by Stempak et al. (95), but rather an adjunct that is most useful when fungi are detected in an anatomic pathology specimen but either they did not grow in culture or a corresponding specimen was not submitted for culture.

Antigen detection tests are available for each of the etiologic agents of the endemic mycoses. Histoplasma antigen assays are commercially available for both urine and serum specimens (96–100). There are conflicting data in the literature regarding the superiority of one of these assays over another, but these authors slightly favor a urine specimen over a serum specimen (101–104). Libert et al. (105), however, demonstrate that either one of these is sufficient in the workup of a patient with suspected histoplasmosis, but ordering both is diagnostically duplicative and increases cost without increasing value. Histoplasma antigen detection testing is most useful for the diagnosis of disseminated histoplasmosis, moderately useful for localized pulmonary histoplasmosis, and not useful for remote disease (97). Antigen detection testing for Blastomyces dermatitidis and Coccidioides species is also commercially available, but it is used considerably less than the Histoplasma antigen assays (106–111). These have been particularly useful for the diagnosis of central nervous system (CNS) disease caused by these fungi (108, 109, 111). Cross-reactivity with substrates from other fungi may occur with these assays and offer opportunities for optimization of future diagnostics (112–114).

Fungal serologic studies are a means of demonstrating that an individual who has the ability to mount an immunologic response has been infected by one of the members of the endemic mycoses. The classic methods used for fungal serology testing are technically complex complement fixation and immunodiffusion (115, 116). The development of low-complexity serologic assays may increase incorporation into clinical laboratory test menus. However, serologic assays for endemic mycoses are of limited to no use during early infection due to the time required for antibody production. In contrast, a rising titer between acute- and convalescent-phase sera is supportive of recent and possibly ongoing infection (115). However, false-positive IgM serology in coccidioidomycosis has been observed in an enzyme immunoassay (117). In addition, serologic tests may be useful when assessing a patient with a solitary pulmonary nodule, particularly when combined with imaging studies and needle biopsy or fine-needle aspiration cytology.

Correlation of direct examination and culture with histopathologic/cytologic findings is a best practice. Enhanced education on organism morphology in tissue and the associated host tissue response would mitigate inaccurate diagnosis and improve patient care. For example, the ability to distinguish H. capsulatum from Candida glabrata based on a neutrophilic (not granulomatous) response, Blastomyces from Cryptococcus based on a pyogranulomatous host response, and Coccidioides endospores from yeasts based on ruptured spherules, granulomas, and eosinophils provides extremely helpful clinically actionable information with significant implications on the choice of antifungal agents and patient morbidity/mortality (118, 119).

The detection and identification of the etiologic agents of the endemic mycoses is important for the accuracy of diagnosis and the direction of antifungal therapy. The combined, but appropriate use of culture, histologic/cytologic studies, antigen testing, serology, and molecular studies afford the timely detection and accurate identification of these important fungal pathogens.

(vi) Fungal infections caused by emerging and underrecognized rare fungal pathogens.

Some fungal organisms are newly emerging or not frequently encountered, but they can cause severe infections and fatal outcomes. Importantly, they pose significant diagnostic challenges to clinical laboratories. The best example of newly emerging fungal pathogen is Candida auris. It is particularly problematic in health care settings due to its multidrug-resistant nature and propensity to cause invasive disease associated with high morbidity and mortality in vulnerable patient populations (120, 121). As hospital systems move toward routine screening for C. auris colonization (122), the inability to correctly and rapidly identify C. auris hinders infection prevention efforts, furthering transmission. While correct identification of C. auris can be done by MALDI-TOF MS (123, 124), misidentifications occur with biochemical reaction-based commercial systems, i.e., Vitek 2, MicroScan, and API tests (123–126). Laboratory-developed NAATs have been successful in the identification of C. auris (127–132), with some methods performed directly on the specimen, reducing turnaround time, and others performed on automated platforms, increasing throughput. We recommend a multicenter approach for developing and evaluating an easy-to-use, cost-effective, rapid, standardized C. auris NAAT directly from the patient specimen.

Rare or uncommon yeasts causing bloodstream infection include but are not limited to uncommon Candida species, Cryptococcus spp. (other than C. neoformans and C. gattii), Trichosporon, Rhodotorula, Malassezia, Geotrichum, and Saprochaete. Most of the bloodstream infections caused by these rare yeasts are either catheter line associated or due to breakthrough on antifungal treatment (133–137). Current automated commercial blood culture systems (BD Bactec Myco/F Lytic bottle, bioMérieux BacT/Alert, and Thermo Fisher VersaTREK) are able to recover these organisms from blood culture except for Malassezia (that requires lipid supplement that is not present in these commercial blood culture systems), but the performance of commercial blood culture systems for detection and isolation of rare yeasts in blood culture is still not well characterized. Current FDA-cleared rapid molecular assays (Biofire FilmArray BCID, GenMark ePlex BCID-FP, and Accelerate Pheno System) for direct identification of yeasts in positive blood culture do not cover these rare yeasts, except for Rhodotorula covered by GenMark ePlex BCID-fungal panel (68). MALDI-TOF MS can also be applied to identify rare yeasts directly on positive blood cultures, e.g., the Bruker Sepsityper (138), but this method may not work well if there is a mixed infection in blood or if that yeast is not in database. If the identification is inconclusive by MALDI-TOF MS, fungal DNA sequencing needs to be performed by targeting the D1D2, ITS, and IGS regions. Determining antifungal susceptibility of rare yeasts is another challenge. Most clinical laboratories rely on commercial antifungal susceptibility testing systems (YeastOne and Vitek), but the performance of these platforms may not be reliable compared to the reference broth microdilution. In addition, there are no Clinical and Laboratory Standards Institute (CLSI) breakpoints or epidemiologic cutoff values (ECVs) to interpret MICs for rare yeast, so little guidance of antifungal drug of choices is provided to clinicians treating patients infected with rare yeasts.

Relatively speaking, any mold infection other than invasive aspergillosis and mucormycosis would be considered a rare mold infection, of which fusariosis and Lomentospora prolificans infection represent the two most important mold infections in immunocompromised patients (139). While isolation of Fusarium from blood or biopsy tissues often supports the diagnosis of the infection, recovery of Fusarium in respiratory samples does not necessarily indicate infection; other laboratory findings and the clinical context of the patient are necessary to make the diagnosis. The serum Aspergillus GM level is often elevated in patients with invasive fusariosis (140, 141). Most Fusarium culture isolates can be reliably identified to species or species complex level by MALDI-TOF MS (142, 143). If these methods are inconclusive, then sequencing identification targeting translation elongation factor-1α region can be applied (144). Isolation of Lomentospora prolificans from non-cystic fibrosis (non-CF) patients with underlying immunocompromised conditions is usually indicative of infection and often results in poor clinical outcome (145). L. prolificans tends to grow slowly on fungal culture media but can be correctly identified by MALDI-TOF MS or microscopic features if the organism sporulates well (142). Since neither a Fusarium- or L. prolificans-specific biomarker nor an FDA-approved NAAT-based molecular assay is available for rapid and early diagnosis of the diseases directly from clinical samples, clinical laboratories primarily rely on culture-based methods to support the diagnosis of these rare mold infections. Such an inadequate diagnostic approach was also seen in diagnosis of other rare mold infections caused by members of dematiaceous fungi (e.g., Alternaria, Bipolaris, Exophiala, Phialophora, Rhinocladiella, etc.), as well as members of hyaline hyphomycetes (e.g., Acremonia, Paecilomyces, Purpureocillium, Rasamsonia, Scopulariopsis, etc.).

Clinical laboratories should be vigilant in detecting any emerging fungal pathogen and recognizing rare or uncommon opportunistic fungal pathogens. A multicenter approach to understand the incidence, clinical features (including risk factors), and antifungal susceptibility profiles of these rare fungal infections is needed. This could be accomplished through a joint effort to develop and validate non-culture-based molecular diagnostic approaches to achieve early and rapid diagnosis of these rare but severe and life-threatening fungal infections.

Method/approach-specific gaps. (i) Direct ID of fungal pathogens in FFPE tissues.

Due to a variety of reasons, tissue for fungal culture is not always submitted with the surgical pathology, leading to all tissues being fixed in formalin and resulting in a diagnosis based solely on histopathological evidence of fungus in formalin-fixed and paraffin-embedded (FFPE) tissue. However, histopathological ID based on morphologic features of fungi (especially molds) is prone to error (146). Lack of culture and erroneous ID based on morphologies on histology render molecular detection of fungi from FFPE tissue particularly helpful in identifying the infectious agent in tissue to the species level (26).

Molecular ID of fungi in FFPE tissue has been successfully achieved by panfungal DNA sequencing ID, targeting ITS, D1/D2, 18S regions (118, 147). Sequencing of hyaline septate molds from FFPE tissue may identify the following clinically actionable items based on species-level IDs: amphotericin B resistance (A. terreus, Scedosporium apiospermum complex, and Purpureocillium lilacinum [formerly Paecilomyces lilacinus]), increased azole resistance (cryptic Aspergillus species like A. lentulus and Paecilomyces variotii), intrinsic voriconazole resistance (Rasamsonia spp.), and resistance to all current antifungal agents (Lomentospora prolificans) (148–150).

Additional benefits to species-level ID by tissue sequencing include the detection of pigmented molds with a predilection to cause CNS lesions (Exophiala dermatitidis, Cladophialaophora bantiana, etc.), ID of the potential source of dermatophyte infections (geophilic, zoophilic, or anthropophilic), and alterations in medical therapy (C. gattii versus C. neoformans, Blastomyces versus unencapsulated Cryptococcus, and Histoplasma capsulatum versus other small yeasts), and tissue processing contaminant versus true infection (151–154).

Although sequencing can identify the isolate to the species level, it also has the potential to detect mutations associated with antifungal resistance, such as Aspergillus CYP51A mutations (155, 156). Novel assays targeting resistance markers and studies generating more species-level AST data correlated with antifungal resistance markers are needed to enhance clinically actionable interpretations of molecular data. Given the expected complexity of such reports, there is also an opportunity for the production of composite infectious disease pathology/microbiology reports to ensure that important test results are not overlooked.

While FFPE tissue sequencing is advantageous for the reasons outlined above, there are important considerations before implementing this assay into routine clinical use. First, protocols for processing, DNA extraction, targeted amplification, sequencing, and bioinformatics analysis are not standardized and vary in their ability to correctly identify fungal organisms (157). Second, fungal databases are limited in comparison to bacterial databases, which can lead to a mis-ID or lack of ID (158). Third, FFPE tissue blocks are processed and handled in a nonsterile manner and thus are prone to contamination. To increase specificity, only FFPE tissue blocks containing histopathologic evidence of fungi should be sequenced.

A targeted next-generation sequencing (NGS) approach has also recently been explored in ID of fungi in FFPE tissue (159). NGS is beneficial for polymicrobial fungal infections, but it is costly and requires highly skilled and trained technologists performing time-consuming manual procedures. Optimizing, automating, streamlining, and standardizing the process would allow more labs to pursue NGS sequencing of FFPE tissue.

(ii) Enhance mold ID by MALDI-TOF MS.

Two MALDI-TOF MS platforms, Vitek MS (bioMérieux, Durham, NC) and Bruker MS (Bruker Daltonics, Billerica, MA), are FDA approved and becoming available in more clinical laboratories. ID of molds using MALDI-TOF MS can be simpler, significantly faster, and more accurate than conventional morphology-based ID. However, it lags behind bacteria and yeasts for utility in clinical microbiology laboratories (142, 160). There are several hurdles to laboratory implementation of MALDI-TOF MS for routine ID of molds. The first is the standardization and expansion of mold databases. Different molds grow better on different culture media, and they grow differently on solid media than on liquid culture. In addition, when molds are grown on plates, unlike most bacteria and yeasts, they have different growth phases and rates that can influence protein expression profiles. For example, a 2-day-old Aspergillus isolate on Czapek’s agar may not give the same profile as the same isolate at 5 days or as a 2-day-old isolate on Sabouraud dextrose agar. Even within the same Aspergillus colony, extraction from the portion predominately containing spores versus the young hyphal mass will generate different proteomic spectra that affects the MADLI-TOF MS ID scores. One way to mitigate this problem is to expand databases to include multiple growth conditions and mold preparations (spores versus hyphae) so that a single validated database can be used for molds grown under diverse conditions. This can be achieved through the sharing of a validated set of mold isolates and the creation of a centralized database to which validated spectra can be added. The other way is to convert molds into hyphal structures by growing them in a liquid culture medium. Once in liquid culture, molds stop sporulation and convert to hyphae within 72 h. The advantage of this method is to eliminate the variation caused by media and to achieve a standard preparation from the hyphal mass, but the limitation is that it adds additional culture time and additional laboratory burden of both time and equipment.

A second significant hurdle to the adoption and expansion of mold ID by MALDI-TOF MS is the variability in each manufacturer’s database. To date, only the Vitek MS mold database has been approved by the FDA for clinical use, while the Bruker Biotyper mold database remains under research-use-only conditions in the United States (161). The number of identifiable organisms in each database differs, with the Vitek MS containing 79 species in the version 3.0 Knowledge Base and the Bruker MS containing 180 species in its V3 library (162). Both of these instruments have shown varying success with the manufacturer-provided databases (161), with Vitek MS users reporting a higher ID rate than those with the Bruker MS. One way to increase successful ID is to supplement the manufacturer’s databases with clinical isolates. These home-grown databases have not only shown improved ID rates, but they also allow for the expansion of libraries by the addition of species that may not be as commonly encountered in some laboratories (163–166). The free distribution of continuously updated and curated databases would provide a significant benefit to clinical laboratories. The creation of MALDI-TOF MS depositories for fungal isolates by groups of clinical laboratories could be the first step in the development of standard databases. These groups could be defined by which instrument they use and could be further refined by geographical area or fungi of interest (e.g., Aspergillus species, dematiaceous molds, etc.). Using a standardized extraction procedure, these groups could produce database updates, provide validation sets containing representative fungal isolates used for database production, and create standardized instructional material, similar to proficiency testing materials.

(iii) Improve detection and isolation of molds in blood culture.

Fungemia, the presence of fungi in the blood, can occur as a result of disseminated fungal infection in patients with malignancies and other forms of immunosuppression (61). Molds are isolated at a much lower rate than Candida but are still important pathogens in this vulnerable patient population since they can have a high rate of mortality (167). The most commonly isolated mold from blood specimens is Fusarium spp. (168), with isolation of Fusarium from 60 to 70% of blood specimens from patients with disseminated fusariosis. Fungemia due to other molds is rare. Aspergillus spp., the most common cause of invasive mold infections, are rarely isolated from blood specimens, with one study finding only 6% of patients with documented invasive aspergillosis had positive blood cultures (169). In addition, as Aspergillus spp. can be common environmental contaminants, determining the clinical significance of a fungal blood culture growing Aspergillus may prove difficult. Other molds, such as Scedosporium apiospermum, Lomentospora prolificans, and Paecilomyces spp., may be emerging causes of fungemia in cancer patients (167) but are rarely recovered from blood culture.

Several commercial blood culture systems have been employed by clinical laboratories to isolate yeast and mold in blood from patients suspected of having invasive fungal infections. These systems include isolator tubes (Wampole Laboratories, Cranbury, NJ), the BD Bactec Myco/F lytic bottle (Becton Dickinson Diagnostic Instrument Systems, Sparks, MD), BacT/Alert (bioMérieux, Inc., Durham, NC), and VersaTREK (Thermo Fisher Scientific). The overall performance data for these automated blood culture systems on recovering molds in blood culture are scant, with Fusarium spp. being the most common. Among these systems, the isolator tube is particularly designed to support isolation of molds. However, its performance on isolating molds is not well known. Of >9,000 pediatric isolator tubes collected from children over a 10-year period, <0.3% recovered a mold (170). The isolator tubes were also prone to false-positive growth due to environmental contamination (171). Although in theory the presence of molds in blood would be inevitable in patients with invasive mold infection due to hematogenous dissemination, it is not clear whether these molds are intrinsically inert to growth in blood or whether these particular commercial blood culture systems are suboptimal to support mold growth in blood. These questions need to be further investigated to find a way to improve the detection and isolation of molds in blood culture. Alternatively, the use of culture-independent methods, including targeted PCR (e.g., aspergillosis PCR), and cell-free NGS are currently emerging as novel approaches to detect molds directly in blood samples and will need further validation (53, 172, 173). Since the incidence of fungemia caused by molds is low in each individual academic center, a multicenter joint effort would be needed in order to optimize and standardize a laboratory protocol for mold blood culture.

(iv) Fungal point-of-care testing.

The immunochromatography technologies have led to the development of lateral flow assays (LFAs) that meet the requirements for point-of-care tests (POCT) for the detection of fungal infections. The POCT are rapid (in minutes), easy to perform, and affordable. One of the best samples is the FDA-approved cryptococcal antigen LFA (IMMY Diagnostics, Norman, OK) with a sensitivity and specificity of >98% in serum and cerebrospinal fluid and a sensitivity of 85% in urine for the diagnosis of cryptococcosis in HIV and non-HIV patients (174, 175). Two commercial LFAs were recently available for the diagnosis of invasive pulmonary aspergillosis (IPA): AspLFD (OLM Diagnostics, United Kingdom), and IMMY sōna Aspergillus GM LFA (IMMY, USA). The AspLFD uses the JF5 monoclonal antibody to detect an extracellular glycoprotein (mannoprotein) antigen secreted by actively growing Aspergillus species, whereas the IMMY GM LFA uses two monoclonal antibodies against Aspergillus GM. Both assays showed a good performance for the diagnosis of IPA in BAL fluids from adult hematology patients and nonneutropenic patients (176, 177). Both LFAs are CE marked (the manufacturer's declaration that the product meets EU standards for health, safety, and environmental protection and indicates that the product may be sold freely in any part of the European Economic Area, regardless of its country of origin) in Europe but not FDA cleared in the United States. A Histoplasma urine antigen LFA has also recently been developed by MiraVista Diagnostics for the rapid detection of disseminated histoplasmosis, showing a sensitivity and a specificity of 96 and 94%, respectively, in patients with AIDS (178). More recently, a Coccidioides LFA assay (IMMY Diagnostics) combined with procalcitonin testing demonstrated correct diagnosis of 77% cases of coccidioidomycosis in a patient cohort from an area of endemicity (179). Overall, the current landscape of the fungal POCT indicates that development of new generation of affordable and rapid LFAs may have the potential to increase clinical laboratory capacity for early and rapid diagnosis of invasive fungal infections.

(v) Optimize and standardize fungal culture procedure for cystic fibrosis patients.

While lung function decline in CF patients has been associated with the recovery of Pseudomonas aeruginosa, Burkholderia cepacia complex, and possibly Staphylococcus aureus in respiratory cultures, the role of fungi in CF pathogenesis is not well understood (180). The American CF Foundation Patient Registry reported in 2019 that prevalence of the most commonly isolated mold, Aspergillus spp. was about 17% in adults (181). However, because the CF Foundation does not explicitly recommend use of selective fungal media in their clinical care guidelines (which are the basis for accreditation of a program as a CF Care Center), it is difficult to know the true prevalence in this population (182). Clinically, recovery of mold in a CF patient’s sputum has been increasingly reported, leading to possible association with decreased lung function, but it is unclear whether mold is a marker of advanced lung disease or a causative agent of lung function decline (180, 183–185).

Most clinical laboratories do not routinely include fungal selective media in CF cultures. Fungi in CF cultures are often recovered from bacterial culture media, which are suboptimal due to lack of antibiotics to suppress bacterial growth and nutrients to support fungal growth. Also, bacterial cultures are typically incubated for shorter periods of time compared to fungal cultures. As a result, slow-growing fungi may be missed. In a recent prospective study, sputum samples from a cohort of CF patients in a single medical center were cultured simultaneously with and without fungal selective media. Inclusion of fungal selective media significantly increased recovery of fungal organisms from 26 to 65% (186). Similar findings were reported by other medical centers (187, 188). Although interest in the role of fungi in pulmonary decline in CF patients is developing, the lack of a widely accepted and standardized CF fungal culture protocol is challenging investigations into the role of fungi.

Development of a standardized and comprehensive fungal culture protocol would require consideration of the following issues. Fungal selective media can be made in-house, but the labor and skill required limit this option to reference laboratories. In contrast, several commercially available fungal media (inhibitory mold agar, Sabouraud agar, and brain heart infusion agar) are readily available and can be easily fit into a clinical laboratory’s workflow. The aforementioned studies confirmed the performance of these commercial fungal selective media to enhance fungal recovery in CF sputum samples (183, 186). In addition, frequency of culture, optimal incubation temperatures and duration, and use of a mucolytic agent during specimen processing may also enhance fungal recovery (187). Multicenter efforts to evaluate optimal culturing practices are needed to develop an evidence-based culture protocol (189). Once a standardized protocol is in place, studies to understand and determine the role of fungi in CF patients (airway colonization versus real infection in conjunction with clinical context) could be further explored.

(vi) Antifungal susceptibility testing.

Antifungal resistance has been reported for all existing antifungal agents in diverse groups of medically important fungi, including Candida and Aspergillus species (190). Although certain species harbor known intrinsic resistance patterns, many fungi exhibit variable resistance necessitating testing of the individual isolate to obtain clinically actionable data. Similar to bacterial AST, CLSI reference methods for fungi include broth microdilution and disk diffusion. However, the available breakpoints are limited to a few of the most common pathogenic yeasts, and only one breakpoint is available for molds (A. fumigatus and voriconazole) (191, 192). Continued expansion of disk diffusion and MIC breakpoints to less-common bug/drug combinations, novel drug classes, and emerging pathogens, such as Candida auris, is essential to improve patient outcome and adapt to new challenges in antifungal resistance (193).

CLSI breakpoint incorporation requires accurate species-level identification paired with MIC, pharmacokinetic/pharmacodynamics, and clinical outcome data. However, given the relatively low frequency compared to bacterial infections, these data are difficult to obtain for less common yeasts and molds. In the interim, yeast and mold ECVs are providing some guidance on AST interpretation for wild-type (WT) and non-WT isolates (194). To move beyond ECVs toward accurate clinically actionable breakpoints will require well-coordinated and longitudinal multicenter studies involving paired clinical outcome data combined with accurate species-level identification and CLSI reference method MIC data provided by specialized mycology reference laboratories.

Although there are several standardized CLSI phenotypic methods available, many clinical labs in the United States do not perform yeast susceptibility testing. Mold susceptibility testing is restricted to a few reference laboratories (195). In contrast to the CLSI broth microdilution reference method, the disk diffusion reference method and commercial gradient diffusion strips are relatively simple (195). There are several commercial automated platforms available for susceptibility testing of yeasts, including Vitek2 and YeastOne (195). Future development of yeast AST on such commercial automated platforms may increase access to many clinical laboratories for in-house streamlined yeast AST, mitigating excess turnaround time and improving time to effective antifungal therapy.

In contrast, there are no commercial platforms for susceptibility testing of molds. While the College of American Pathologists offers proficiency testing for yeast susceptibility testing, it does not offer proficiency testing for mold susceptibility testing. For these reasons, mold susceptibility testing is offered only by a few large reference laboratories in the United States, with relatively long turnaround times. More commercial assays for fungal susceptibility testing, especially platforms that support mold susceptibility testing, are needed. Multicenter studies aimed at identifying the accuracy and clinical utility of in-house commercial gradient diffusion strips and other commercial platforms versus send-out testing are merited. Since three new classes of mold-active antifungals are currently in development, there may soon be multiple good choices for the treatment of mold infections. In this context, mold antifungal susceptibility testing will be essential for maximizing effective therapy (196).