Despite the availability of new antifungal drugs, the overall survival for immunocompromised patients with invasive fungal infections remains too low, with large variations according to underlying disease.1–15 Although early diagnosis and subsequent early initiation of therapy improves outcome,16–19 diagnosing invasive fungal infections can be difficult. The purpose of this article is to review the available armamentaria for the diagnosis of invasive fungal infections. A brief summary of the main clinical and epidemiological data for these infections is shown in Table 1.3–7,12–15,20–56
Diagnosis of Invasive Fungal Infection
Conventional Methods (Direct Microscopy, Culture and Histopathology)
All fungi obtained from sterile sites should be identified to species level by referral to a specialist laboratory. All bronchoscopy fluids from patients suspected of infection should be examined microscopically for hyphae and cultured on specialised media, and all clinical isolates of Aspergillus should be identified to species level.57
Current ‘conventional methods’ are very limited for the diagnosis of invasive fungal infections. Blood cultures have a low sensitivity for the diagnosis of candidaemia (~50%),58,59 and cultures other than blood are non-specific and can take too long to became positive. Antifungal treatment is recommended following recovery of even one positive blood culture for Candida.5 Identification of Candida spp. by culture requires the presence of viable organisms in blood or body fluids. In addition, several days may be required for blood cultures to become positive and, for non-Albicans spp. of Candida, additional subculturing is required to obtain pure cultures for use in subsequent phenotypic identification systems.59
Although the lungs are frequently involved in disseminated candosis, primary Candida pneumonia is a rare condition,60,61 and benign colonisation of the airway with Candida spp. and/or contamination of the respiratory secretions with oropharyngeal material is much more common than true Candida pneumonia. Thus, diagnoses of Candida pneumonia that are based solely on microbiological data are often incorrect. In addition, debate persists about the significance of the isolation of Candida in the peritoneal fluid,62 and the presence of Candida in the urine usually represents colonisation, despite its presence in 9% of hospitalised patients in the US.63
For the diagnosis of invasive aspergillosis, cultures of the respiratory tract secretions lack sensitivity. Aspergillus is grown from sputum in only 8% to 34%, and from broncho-alveolar lavage (BAL) in 45% to 62% of patients with invasive aspergillosis.64 Aspergillus recovery from the respiratory tract usually represents colonisation in immunocompetent patients but may strongly suggest invasive disease in the immunocompromised host.65,66 While confirmation of the diagnosis of invasive aspergillosis has typically required histopathologic evaluation, profound neutropaenia and thrombocytopaenia often preclude the pursuit of biopsies. Transbronchial biopsy or brushings are too often false negative. Biopsies of endobronchial lesions have been useful when such lesions are encountered. Blood, cerebrospinal fluid (CSF) and bone marrow specimens rarely yield Aspergillus spp.33
In contrast to disseminated aspergillosis, disseminated fusariosis can be diagnosed by blood cultures in 40% of patients.12,15 The rate of positive blood cultures increases to 60% in the presence of disseminated skin lesions.14 Microscopically, the hyphae of Fusarium in tissue resemble those of Aspergillus; the filaments are hyaline, septate and 3–8μm in diameter. They typically branch at acute or right angles. The production of both fusoid macroconidia and microconidia are characteristic of the genus Fusarium.13 Skin lesions should be submitted to biopsy.
The diagnosis of zygomycosis is usually made histologically, and the demonstration of fungal elements from cytologic preparations is complicated by the difficulty of extracting fungal elements from invaded tissues.11 The poor sensitivity of sputum culture (<25%) makes diagnosis of pulmonary zygomycosis challenging.38 The yield of BAL is not higher,38,67 but direct microscopy of BAL together with transbronchial biopsy may increase the yield.43 A positive finding from BAL from a neutropaenic or immunocompromised host would be highly suggestive of infection, and should be treated as such.43 Even though, on microscopy, Mucorales have been classically described as having broad (10 to 50μm), ribbon-like aseptate hyphae with right-angle branching, the hyphae are actually pauciseptate, and the angle of hyphal branching can vary from 45° to 90°, reinforcing the importance of obtaining material for culture.68 About 80% of disseminated infections with S. prolificans are associated with positive blood cultures, but this proportion is much lower with S. apiospermum infections.45
All tissues from patients with suspected infection should be stained with fungal stains in parallel with regular stains. The practice of assessing H&E stains of tissues before deciding whether to use specialised stains for fungi frequently introduces fatal delays for patients. Reporting of specimens containing any fungal elements should always include the presence and absence of yeast forms, hyphae and whether or not they are septate, if it is possible to tell, and whether there is any melanin present.57 However, while the demonstration of Aspergillus in tissue is the reference standard for diagnosis of invasive aspergillosis, a definitive diagnosis is possible only after identification of the fungus cultured from that tissue.69 So, part of the biopsed tissue or other surgical specimen should be sent to the microbiology laboratory (not in formalin), in addition to the pathology laboratory. Immunohistological staining using polyclonal fluorescent antibody reagents can distinguish Aspergillus spp. from Fusarium spp.70 In situ hybridisation may also help to distinguish Fusarium spp. from Aspergillus and Pseudoallescheria in tissue sections.71
New Diagnostic Tool
Galactomannan
Galactomannan is a cell wall polysaccharide released by Aspergillus spp. during fungal growth in tissue.69,72,73 A commercially available sandwich ELISA (Platelia Aspergillus, BioRad) detects galactomannan by use of a rat monoclonal antibody. The test has been validated for serum specimens only and has a detection limit of ~1ng/mL, which is 10 to 15 times lower than the limit of the latex agglutination test used previously (Pastorex Aspergillus, Biorad).73–76 Circulating galactomannan may be detected at a median of five to eight days before clinical manifestation of aspergillosis.72,77–80 The concentration of circulating galactomannan corresponds with the fungal tissue burden81,82 and may therefore be used to monitor the response to treatment.77,78
Studies evaluating the role of galactomannan assay in the diagnosis of invasive aspergillosis have largely been conducted with leukaemia patients or haematopoietic stem-cell transplantation (HSCT) recipients.77,78,82–89 After initial clinical studies suggested a high sensitivity and specificity,73,76,78,79,90,91 further studies revealed high rates of false-positive results among paediatric patients and neonates.81,92,93 In one study, rates of false-positive results as high as 83% were observed93 that may be related to cross-reactivity with Bifidobacterium bifidum, found in large inocula in the guts of breast- and formula-fed infants.94 The presence of a damaged gut endothelium may increase the absorption of dietary galactomannan.95 Specificity was also lower in adult allo-HSCT recipients than in adult auto-HSCT recipients or non-transplant patients.92 The rate of false-positive results is high in the first 30 days following bone marrow transplantation and 10 days after starting cytotoxic chemotherapy.96,97 The use of galactomannan as a surveillance tool in transplant recipients has been associated with a positive predictive value of only 10%.92 The results also suggest that routine ELISA tests are not useful in patients with febrile neutropaenia with no clinical or radiological signs suggestive of a pulmonary infection.92 Otherwise, all high-risk patients with a respiratory tract infection or suspected extrapulmonary aspergillosis should be repeatedly tested with galactomannan ELISA, as the predictive positive value of the assay was highest in these groups.92
Cross-reactivity of Platelia Aspergillus galactomannan with Penicillium spp. has been noted96 but is deemed to be of little clinical relevance since Penicillium spp. are rarely pathogens in humans. In addition, drugs of fungal origin, such as antibiotics, may be associated with a false-positive test, including ampicillin-sulbactam, piperacillin-tazobactam, and amoxicillin-clavulanic acid.98–102 The timing of collection of the sample may influence the test results, with reactivity being less likely in samples collected at trough levels or prior to the administration of the dose.103 ELISA cross-reactivity has been observed with other fungi, including Paecilomyces variotii, and Alternaria spp.92 False-positive reactions have also been seen in bacteraemic patients,96 and in those with autoreactive antibodies.64,88
In a study of 3,924 serum samples in cancer patients, the overall sensitivity of the ELISA galactomannan test was only 29.4% (64.5% when only patients with proven invasive aspergillosis were analysed).92 Low sensitivity for the ELISA assay has also been reported in allo-HSCT recipients (60%), lung transplant recipients (30%),89 liver transplant recipients (56%), and in patients with various other conditions including non-malignant diseases (52%).83,93 While a positive test in a lung transplant recipient with a clinical illness compatible with invasive aspergillosis may be considered highly suggestive of this infection, a negative test does not rule out aspergillosis.64 The sensitivity of the test has been typically lower in non-neutropaenic patients (15% to 30%) and may be related to lower circulating galactomannan levels.64 The use of both prophylactic and empiric antifungals may also lower antigen levels by decreasing the fungal load.64
Overall, the galactomannan test seems to be a highly specific diagnostic tool (94% to 99%), even though sensitivity has ranged from 50% to 93% in patients with haematologic malignancy.72,77,78 In order to improve these results, some authors have suggested that the recommended cut-off value of the test should be reduced from 1,500 to 1,000.72,77 A further reduction to 0.700 was suggested for adults who have not undergone allogeneic transplantation,92 and recent research suggests that an index of 0.5 is a more definitive threshold.104 These modifications may increase the sensitivity of the test, with only a low decrease in specificity. Discussion persists about the best cut-off for galactomannan assays.
Because galactomannan is a water-soluble carbohydrate, it can be detected in other fluids.69 Although the antigen can be detected in the urine,73,74,105–107 little is known about its pharmacokinetics and clearance by the kidney. The effect of renal failure or dialysis on the clearance of galactomannan is also indefinite. Little is known regarding the correlation between galactomannan detection in urine and disease progression, and false-positive results may be an important drawback for this test.69,73,74
Because galactomannan is predominantly released by Aspergillus hyphae during growth and to a much lesser extent by conidia, detection of galactomannan in BAL fluid provides better evidence for aspergillosis than culture108,109 or polymerase chain reactions (PCRs), which do not discriminate between contamination conidia and hyphae.110–112 However, false positive results may occur in patients only colonised with Aspergillus when BAL is tested,69 but this can be improved when one combines this method with the high-resolution computed tomography (CT) scan.34,113
Diagnosis of CNS aspergillosis is very difficult and even brain biopsies do not always result in a clear diagnosis.69 Culture of CSF infrequently yields positive results,94,114 and both chemical findings in the CSF and the CT results are often non-specific.115–117 Galactomannan may also be detected in the CSF, with a sensitivity and specificity of 80% and 100%, respectively, according to a study involving only five patients with proven CNS aspergillosis.117 Galactomannan has been detected in other clinical specimens as well.69
1,3-beta-D-glucan
1,3-beta-D-glucan is a cell wall component of yeast and filamentous fungi, which is detectable in the blood during most invasive fungal infections. Factor G, a coagulation factor, is a sensitive natural detector of this antigen.118 The reported sensitivity and specificity for the assay have ranged from 67% to 100%, and 84% to 100%, respectively.118–122 The test does not detect cryptococcosis, and it is also not positive in cases of oral candidosis or fungal colonisation.121 However, in addition to invasive aspergillosis and candidosis, it detects infections caused by species of Fusarium, Trichosporon, Saccharomyces and Acremonium, which are less common but equally important fungal pathogens, especially in immunocompromised hosts.121 False positive tests have been reported in patients undergoing haemodialysis, patients with cirrhosis, recipients of antitumour polysaccharides, and patients immediately following abdominal surgery.64
PCRn
Although PCR is at lest 19 times more sensitive than culture for A. fumigatus,123 PCR-based molecular diagnostic tests for aspergillosis are not commercially available and remain largely unstandardised. A sensitivity of 79% to 100% and a specificity of 81% to 93% have been documented, depending on the methodology used.124–126 Such assays, when performed on blood or BAL samples, have shown a negative predictive value for invasive aspergillosis ranging from 92% to 99%. However, PCR-based assays performed with BAL samples have shown low positive predictive values that are likely to reflect respiratory tract colonisation.127
Markers for Invasive Candidosis
Mannan is a cell wall surface carbohydrate that circulates during infection with Candida spp.,128 and studies have shown a correlation between detectable mannanemia and tissue invasive in patients with candidaemia.129 However, mannan is rapidly cleared from the blood and occurs in low levels, necessitating frequent sampling for detection.130 In addition, this is a very expensive test. D-arabinitol is a metabolite of certain species of Candida that accumulates in the urine of patients with invasive candidosis. Assay sensitivity for D-arabinitol is only ~50%, and does not detect C. krusei and C. glabrata.131,132 Enolase is perhaps the antigen with the greatest promise for the diagnosis of invasive candidosis.133,134 The sensitivities of assays for enolase have ranged from 54% to 75%, and higher sensitivity may be achieved with serial testing. Furthermore, the enolase antigen is highly specific for Candida spp., and is not present in superficial Candida colonisation.133
Experimental models and clinical studies have shown PCR to be more sensitive than culture for detection of candidaemia. The sensitivity of the Candida PCR assay was 95.0% in one study, compared with a sensitivity of 75.0% for the Candida ELISA aiming to detect mannan (a difference not statistically different).104 The specificities of the Candida PCR and ELISA were the same at 97.0%. In 45% of these patients, the PCR method detected the infection earlier than the ELISA. Newer realtime PCR assays such as TaqMan and the Light Cycler require no post-amplification manipulations and can potentially be automated for all steps from DNA extraction to final PCR amplicon detection and quantitation.59
It should be noted that there have been several headto- head comparisons of the various assays for detection of the Candida antigens discussed earlier. Unfortunately, none of the assays have performed well enough or has good enough predictive value at this point to be able to recommend its routine use in a clinical laboratory. Perhaps a combination of two assays may increase the accuracy of diagnosis of invasive candidosis.130
Markers for Other Invasive Fungal Infections
The amplification of gene sequences unique to fungi is conceptually appealing, offering the potential for rapid and sensitive diagnosis of invasive fungal infections. In general, assays targeting multicopy genes have better detection limits than those targeting single copy genes.135 Although PCR may become a diagnostic modality to identify Mucorales,136–138 Scedosporium,139–141 and several other fungi,142–151 no commercial test is available to date. These techniques hold promise, but they are not yet standardised or readily available in most clinical laboratories. Also, large clinical trials to determine the sensitivity and specificity of such molecular tests are non-existent.
Role of the CT Scan
The ‘halo sign’ refers to a zone of ground-glass attenuation surrounding a pulmonary nodule or mass on computed tomography (CT) images.152 The presence of a halo of ground-glass attenuation is usually associated with haemorrhagic nodules.153 In severely neutropaenic patients, this suggests infection by an angioinvasive fungus, most commonly Aspergillus. The halo sign is documented in 33% to 60% of patients with invasive aspergillosis and is short-lived.154 To be useful for the diagnosis of aspergillosis, the CT scan must be performed within five days of the onset of infection, because ~75% of the initial halo signs disappear within a week.155 The ‘air crescent’ sign does not appear until the third week of the illness, and its appearance may be too delayed to be helpful in the diagnosis of invasive aspergillosis.154
Although usually regarded as an indication of haemorrhagic nodules, the halo sign may also be present when tumour or inflammatory cells infiltrate the lung parenchyma.153,155–159 The halo has been described in patients with eosinophilic pneumonia, bronchiolitis obliterans organising pneumonia,160 and tuberculosis,161 and in patients infected by Mycobacterium avium complex,155 Coxiella burnetti,162 cytomegalovirus, herpes simplex virus,153 and myxovirus.160 Patients with post-transplantation lymphoproliferative disorders163 and Wegener granulomatosis may develop the halo sign,153 as well as some of those who have undergone transbronchial biopsy.155 Although it is less common, the halo sign may also be observed in non-haemorrhagic nodules, in which case either tumour cells or inflammatory infiltration account for the halo of ground-glass attenuation.153 Nonetheless, in the appropriate clinical setting, the halo sign is considered to be early evidence of pulmonary aspergillosis even before serologic tests become positive,164 and it warrants the administration of systemic antifungal therapy.165
Future Perspectives
Invasive fungal infections constitute a major challenge for the management of immuno-compromised patients, mainly haemato-oncology patients, transplant and allo-HSCT recipients. In order to improve the survival for these infections, an early diagnosis is required. As shown in this article, conventional microbiological, histological and radiological techniques remain the cornerstone of diagnosis but are insensitive and have a limited impact on clinical decision-making. There is an urgent need for new and efficient diagnostic methods. These tests should be fast and highly sensitive. In addition, the recognition of the causal agent should be very precise. With the advance of molecular tools, new fungal species and new mechanisms of resistance will be clarified. DNA-and RNA-based methods hold promise for improved sensitivity and specificity, but these methods will require extensive validation in clinical studies. Finally, costs are also very important in selecting an appropriate diagnostic test for invasive fungal infections. ■