There are 4 antigen tests available with excellent to reasonable clinical performance. Antigen testing is directed against a fungus-synthesised and released carbohydrate into body fluids. These antigens are generally poorly immunogenic and so not usually heavily complexed with specific antibody, with the exception of the Candida mannan antigen.
Cryptococcal antigen testing
The glucuronoxylomannan long branched capsular polysaccharide antigen of Cryptococcus species is released in abundance in body fluids by all common pathogenic species. Multiple different detection systems have been detected often abbreviated as CRAG (CRyptococcal AntiGen test). The most widely used has been the latex agglutination test in which serum or CSF can be tested. Several companies make CRAG tests including Meridian Diagnostics, International Biological Labs, MicroScan, Eiken, Murex Biotech and (Immy). Usually a screening test at neat and a 10-fold dilution is done as a screen. If positive 2-fold dilutions are then done to determine a titre. The prozone phenomenon was described with the CRAG latex in testing serum and so the serum is first boiled and pronase treated, in most kits, to abolish this problem as well as releasing any bound antigen. Recently a new dipstick point-of-care test (Immy) has been introduced avoiding this serum treatment step and also enabling testing of urinary cryptococcal antigen, which is useful for screening. Some labs have tested respiratory fluids for cryptococcal antigen to diagnose cryptococcal pneumonia.
A recently published study (2015) with 207 patients, comparing CRAG (IMMY) tests performed on fingerstick whole blood, with CSF and plasma/serum, showed 100% agreement of CRAG-LFA performed on fingerstick blood with other body fluids - suggesting that fingerstick CRAG is a reliable bedside diagnostic test. Lateral flow assay is a 10 minute procedure and can be done by almost any personnel. There are 5 kits on the market: Immy, Era Biology, Bio-Rad; Liming BIO and Biosynex ). They all have excellent performance. The Biosynex LFA uses a reader to allow the result to be included in the lab IT system and provides slightly higher sentivity. There are also 2 bands on the Biosynex and the presence of a higher concentration of antigen in the blood correlates well with the presence of cryptococcal meningitis.
This video demonstrates how to undertake CRAG LFA with a fingerstick blood sample (courtesy of Nathan Yueh).
In HIV/AIDS, the cryptococcal antigen is positive in CSF in >99% of cases of cryptococcal meningitis and almost all of these patients also have detectable serum antigen positives. Using the new dipstick antigen test, almost all are also positive in urine, although the amount of antigen detectable is 20-fold lower. Serum and possibly urine screening may identify cryptococcal disease early at a less advanced state and may reduce the number of cases of meningitis or the seriousness of the infection.
In other immunocompromised patients with cryptococcal meningitis, the serum assay is negative in about 30% . CSF titres tend to be lower to those of HIV/AIDS patients with meningitis.No data are published for urine detection.
In non-immunocompromised patients, with cryptococcal disease (pneumonia or meningitis) the serum CRAG is often negative, and only the CSF or BAL positive, usually at low titre. Often C. neoformans is not recovered from clinical samples, a positive CRAG is the only diagnostic data.
The CRAG titre is useful for following response to therapy in the CSF, but not so useful in the serum. Titres fall slowly and variation in testing may not give a uniform fall, unless prior samples are tested together with the current sample. (Comparison of commercial kits). Usually 2-fold dilution titres are done with latex agglutination, but there are also ELISA kits with a quantitative endpoint (Meridian, Immy and Dynamiker.)
CrAg Lateral Flow Device for the detection of cryptococcal antigen
Aspergillus antigen testing
Galactomannan is secreted by all pathogenic Aspergillus species, although some isolates are poor producers. Three commercial assays are available; Platelia Aspergillus EIA BioRad, using an ELISA format, the Dynamiker Aspergillus Galactomannan Assay, by Dynamiker, and the Goldstream Aspergillus Galactomannan ELISA Test, by ERA Biology. A similar metabolite is also produced by several other fungal species including Penicillium spp. Histoplasma capsulatum and other fungi.
Two lateral flow assays (dipsticks) are available for invasive aspergillosis, from IMMY and ISCA Diagnostics (distributed by OLM Diagnostics). These can be performed on either serum or BAL and give results in around half an hour.
Sample type and cutoff
The assay was originally designed for serum, but has also been used on bronchoalveolar lavage fluid (BAL), sputum, CSF, pleural fluid, pericardial fluid and tissue. The generally accepted cutoff in serum for positives is 0.5. The higher the OD the greater the likelihood of invasive aspergillosis. No cutoff has been defined for any other specimen type, but in BAL fluids different authors have recommended cutoffs from 1.0 to 2.0. This variation partly reflects different patient populations and different dilution factors during bronchoalveolar lavage. Pretreatment of BAL fluid with Sputasol (dithrietol) greatly reduces galactomannan levels.
Specificity and false positive results
Overall specificity of Aspergillus antigen detection using the galactomannan EIA is approximately 80%. In addition to a lack of specificity with respect to some rarer fungi (in which case probably the patient will receive antifungal therapy), several antibiotics and other patient factors may yield false positives. Several beta-lactam antibiotics are manufactured using fungal fermentation as s first step. This often results in carryover of galactomannan into the antibiotic preparation and ‘false positive’ galactomannan results that can persist for some days. The most common antibiotics to be implicated are piperacillin/tazobactam (Tazocin) and amoxicillin/sulbactam (Augmentin), but some cephalosporins and meropenom have also been implicated. False positive results have not been observed with glycopeptides, quinolones or aminoglycosides. In addition to this, many foods contain galactomannan including pasta and yoghurt (View false positive data). In patients with permeable small bowel, such as those with mucositis after chemotherapy, false positive galactomannan tests may occur, and perhaps slightly more commonly in children.
Sensitivity and false negative tests
The sensitivity of Aspergillus antigen detection using the galactomannan EIA in serum in neutropenic patients not receiving itraconazole or posaconazole prophylaxis (and possibly other agents) is approximately 80%. Itraconazole and posaconazole antifungal prophylaxis reduce the sensitivity to 0-20%. In addition, some isolates of Aspergillus appear to be poor producers of galactomannan. While galactomannan usually is detectable up to a week before a CT scan or other standard diagnostic tests are positive, in some cases, it is produced late in infection.
In non-neutropenic patients, reduced circulating and possibly antigen complexing reduces the serum sensitivity to 0-25%. However, respiratory samples may still be positive. For ventilated intensive care patients with invasive aspergillosis, galactomannan is detectable in ~85% of BAL samples and is the best means currently of establishing a probable diagnosis. Conversely, positive BAL galactomannan tests may be converted to negatives within 3 days by antifungal therapy, so false negative results should be expected soon after anti-Aspergillus therapy is started.
Response to therapy
In haematology patients (profound neutropenia and haematopoietic stem cell transplantation), failure of therapy is associated with persistently elevated galactomannan in serum and conversely a fall in OD by 3 weeks is associated with good outcomes.
Histoplasma antigen testing
As Histoplasma capsulatum is slow growing (10 days or more), early diagnosis of disseminated infection required visualisation of the organism in blood smears, bone marrow, lymph node, skin or lung biopsies. Several antigens are detected, including some carbohydrate or glycoprotein structures. The optimal specimen is urine, although blood and BAL fluids have also been tested. Histoplasma antigen testing is now on the WHO Essential Diagnostics List (read a report published after the GAFFI meeting calling for its inclusion).
Two tests are commercially available in ELISA format. One has a sensitivity of 81% and a specificity of 99% and a diagnostic accuracy of 96% (Immy). There is cross-reactivity with Blastomyces dermatitidis, Coccidioides immitis and Paracoccidioides brasiliensis. View Immy test data. The other has a sensitivity of 95% (100% against culture positives) and a specificity of 70% with a diagnostic accuracy of 86%. (Optimum Imaging). There is cross-reactivity with Blastomyces dermatitidis, Coccidioides immitis and Cryptococcus neoformans.
Another reference test is also available as a service for specimens submitted (Miravista Labs) and this assay performs well, with well characterised specimens. Antigenuria was detected in 92% of those with disseminated histoplasmosis and antigenemia in 100%. Those with severe disease and immunocompromised patients are more often positive. Lower frequencies of positivity are found in those with acute, subacute and chronic pulmonary histoplasmosis. Cross-reactivity was common in patients with blastomycosis.
A recent publication has evaluated the MiraVista EIA tests.
A mannan antigen of Candida albicans circulated in the blood of many patients with candidaemia. It is immunogenic and antibodies are formed over time. The relatively low sensitivity is explained by this antibody complexing. Overall diagnostic performance is improved by also measuring anti-mannan antibodies. Several test kits are available with differing performance.
The sensitivity and specificity of the Cand-Tec kit (Ramco) were 52.6 and 50.5% respectively in patients with candidaemia.
The sensitivity and specificity of the Cica Fungi Test (Kanto Chemical) were 63.2 and 95.5% respectively in patients with candidaemia.
In invasive candidiasis, most of whom had candidaemia, mannan detection (BioRad) sensitivity was 58% and specificity 93% with a diagnostic odds ratio (DOR) of 18. Concurrent detection of anti-mannan antibodies increased the performance of the test to a sensitivity of 83% specificity of 86% and DOR of 58. In neutropenic patients, at least one positive mannan or anti-mannan test is positive, and so interpretation is improved by repeat testing and requiring 2 positives for a true positive.
Testing for aspergillus antigens
Tecan plate reader for ELISA assays