Antibody detection assays were the area of earliest interest but yielded tests with poor sensitivity and specificity. These tests have since fallen out of favor and are not recommended for routine use.

Several tests that target a variety of cell wall and cytoplasmic components have been developed to detect macromolecular Candida antigens (Table 5.2). Some of these are no longer available, such as an assay to detect enolase antigen. Of the available tests, the earliest was the Cand-Tec LA assay (Ramco Laboratories, Stafford, TX), which was designed to detect circulating Candida antigen in patients with serious, disseminated infection. Unfortunately, there are conflicting reports on its overall sensitivity and specificity, especially in patients with renal failure or rheumatoid factor positivity, making it difficult to confirm the diagnosis of candidiasis by the Cand-Tec assay alone [20, 21].

The mannan component of the Candida cell wall is a major antigen and the target of many serum detection assays. These assays vary in the laboratory method and type of antibody used for antigen detection. Two tests that use the same monoclonal antibody are the Pastorex Candida LA test (Bio-Rad, Marnes-la-Coquette, France) and the Platelia Candida Antigen EIA test (Bio-Rad, Marnes-la-Coquette, France). Although the EIA test is more sensitive than the LA test, they are both limited by the rapid clearance of mannan antigenemia. In an effort to overcome this, an anti-mannan antibody EIA was developed and marketed individually (as the Platelia Candida Antibody test) and combined with the Platelia Candida Antigen (as the Platelia Candida Antibody and Antigen test) [22]. In a review of 14 studies that evaluated the three Platelia Candida assays, the pooled per-patient sensitivity of the Platelia Candida Antigen, Platelia Candida Antibody, and both tests combined was 58, 59, and 83 %, respectively, while the corresponding specificities were 93, 83, and 86 %, respectively [23]. Both Platelia Candida tests have been recently refined and are now marketed as Candida Ag Plus and Candida Ab Plus. In one initial evaluation of their performance using a case–control design, their sensitivity and specificity were, respectively, 59 and 98 % for Candida Ag Plus, 63 and 65 % for Candida Ab Plus, and 89 and 63.0 % for the tests combined (Candida Ag/Ab Plus) [24]. Further evaluation of the newer test is warranted before widespread clinical application.

D-arabinitol (DA) is a five carbon polyol metabolite that is produced by several pathogenic Candida species (except for C. krusei and C. glabrata). It has been shown to be present in higher serum concentrations in humans and animals with IC than in uninfected or colonized controls, making it potentially useful as a diagnostic marker for IC. There are two general methods to measure DA: gas chromatography or an enzymatic method. The former is labor intensive and not readily available in most hospital laboratories, while the latter is more suited to a commercial test kit, as is currently marketed in Japan as Arabinitec-Auto (Marukin Diagnostics, Osaka, Japan). This assay is also available for DA testing on urine samples. Several studies have shown that DA can be detected earlier than the presence of Candida in blood cultures and that serial measurements correlate well with clinical response to therapy [25, 26].

Immunodiagnostic tests are promising additional diagnostic strategies in the detection of IC, but they have yet to supplant traditional methods. In general, there appears to be utility for serum BG, serum mannan antigen/antibody, and serum DA as adjunctive tests in combination with cultures, histopathology, and radiology. As yet, no single test has been demonstrated to have optimal sensitivity and specificity. Their greatest value appears to be in serial testing of high-risk populations where the trend (rather than a single value) will allow an earlier, accurate diagnosis and monitor the effectiveness of empirically instituted antifungal therapy. An emerging strategy is to use these tests in combination with one another and/or molecular assays to optimize a diagnostic approach to these complex patients. Further, rigorous, prospective clinical trials are needed to determine which tests or combinations thereof will offer the greatest clinical utility.

Galactomannan (GM) is a polysaccharide component of the Aspergillus cell wall. It has been demonstrated in the serum of some patients with IA and thus has been the target of several serum detection assays. An earlier test called the Pastorex Aspergillus (Bio-Rad, Hercules, CA) utilized an LA method with a monoclonal antibody. This test yielded disappointing results with low sensitivity unless multiple samples were used and false-positive reactions from cross-reactivity of the antibody with several other fungal species (see Table 5.3) [27, 28]. A newer, commercially available test is the Platelia Aspergillus Ag (Bio-Rad, Hercules, CA). This sandwich ELISA uses the same monoclonal antibody but has the ability to detect GM at much lower limits, thereby improving the test’s sensitivity and allowing earlier detection of IA. In a retrospective review of stored serum specimens on bone marrow transplant and leukemia patients, the study leading to its FDA approval, this assay had a sensitivity of 81 % and specificity of 89 % [29]. However, subsequent investigations have shown a lower sensitivity; for example, a meta-analysis of 27 studies of the Platelia Aspergillus assay in serum, performed in the setting of repeated surveillance in high-risk patients, demonstrated an overall sensitivity of 61 % and specificity of 93 % in proven or probable IA cases [30]. The sensitivity of the test, when limited to proven cases only, was 71 %. In subgroup analyses, the sensitivity was 79 % in those studies which reported using an optical density (OD) cutoff at the US FDA-cleared threshold of 0.5 (as opposed to a higher OD cutoff of 1 or 1.5; n = 5). GM test performance is best in the hemato-oncologic population and is less sensitive in solid organ transplant patients [30]. An emerging role for serum GM testing is to monitor treatment response; several recent reports have demonstrated that the trend of GM during therapy predicts outcome [31–33]. False-negative results can occur due to limited angioinvasion, low fungal load, high antibody titers, or the use of prophylactic or preemptive antifungals. Alternatively, false positives occur due to cross-reactivity of the assay with other fungal species, co-administration of pipercillin-tazobactam, or several fungal-derived antibiotics.

The Platelia Aspergillus assay was FDA-cleared for use on bronchoalveolar lavage (BAL) fluid in 2011, and its use has increased due to the improved sensitivity (generally > 75 %) as compared to serum across various patient populations [34–36]. A meta-analysis of 30 studies demonstrated a sensitivity of 87 % and specificity of 89 % for the BAL-GM [37]. In a systematic review of seven studies comparing BAL GM (OD cutoff 0.5) to Aspergillus PCR, BAL GM performed similarly; BAL GM sensitivity and specificity were reported as 82 and 97 %, respectively [38]. As with serum GM, test performance depends on the selected OD cutoff required for test positivity (FDA-cleared at 0.5). While specificity is improved, sensitivity is sacrificed if higher OD cutoffs (e.g., 1.0 or 1.5) are used to indicate a positive result. Though it can be detected in other body fluids, the GM assay is validated only for serum and BAL samples at this time.

Despite the broad spectrum of disease caused by Aspergillus organisms, it is the invasive disease that is most important and most challenging diagnostically. The Platelia Aspergillus assay supports a diagnosis of IA in the appropriate clinical setting. This serum test has shown increased specificity with serial sampling and is ideally used to screen patients on a weekly or twice weekly basis during periods of severe immunosuppression and may be useful to monitor patients on therapy. All clinicians should keep in mind the potential for false positive and negative results and incorporate the GM results into the general clinical assessment of the patient, rather than as the sole basis on which to change management. For instance, a change in the assay from negative to positive in an immunosuppressed patient under surveillance should prompt a more thorough investigation for IA, while a change from positive to negative should lend support to other evidence that proper therapy has been instituted. Additionally, it is important to remember that the positive predictive value of this test is highest in populations with a high pretest probability; using it for routine diagnosis in lower risk populations will likely increase the chance that a positive result is a false positive. Also, BAL fluid GM testing demonstrates very good sensitivity and specificity although similar cautions regarding test interpretation should be considered. Despite its limitations, this assay is a suitable, noninvasive adjunct for diagnosing and managing IA (see Box 5.2 for recommendations).

Tests for cryptococcal antibodies are not useful and are not widely available for clinical use because they have high false positive and false negative rates.

Some of the most important and rapid serodiagnostic tests available for any fungi are those used to detect the cryptococcal capsular polysaccharide antigen, glucuronoxylomannan (GXM) (Table 5.4). These tests utilize a variety of different laboratory techniques (e.g., LA and EIA) for antigen detection. Recently, a point-of-care LFA (IMMY, Norman, OK) was FDA-cleared in 2011. In several studies, the available assays have been directly compared; assay performance is generally excellent in serum (with some important inter-assay differences) and outstanding in CSF (with minimal inter-assay differences) [40–44]. Potential advantages of the LFA are that it is easily stored (room temperature), rapid, and simple to perform/interpret. However, the LFA requires further evaluation and particularly validation of quantitative cryptococcal antigen titers before its widespread adoption.

The performance characteristics of the cryptococcal antigen assays to diagnose cryptococcosis depend on both the disease status (localized vs. disseminated) and host status (HIV, transplant, otherwise immunosuppressed, or immune competent). Due to the multiple possible combinations, the ability to tease out these performance differences is difficult. In general, a higher fungal burden (as is typical for CNS disease) is associated with higher antigen titers. For example, the sensitivity of serum cryptococcal antigen is lower for isolated pulmonary disease than that for meningitis. Indeed, in immunosuppressed patient, a positive serum cryptococcal antigen from a patient who may only appear to have localized pulmonary disease should prompt the clinician to strongly consider checking blood cultures and CSF fluid to rule out dissemination [45]. There has been much investigation regarding the differing performing characteristics of the ability of serum and CSF antigen tests to detect cryptococcal meningitis. In a review of nine studies, the sensitivity of serum and CSF antigen titers to detect CNS disease in the following patient groups was reported: 92 % (307/333) and 99 % (181/183) in HIV-positive patients; 88 % (38/43) and 100 % (65/65) in transplant recipients; 87 % (91/105) and 97 % (141/154) in otherwise immunocompromised patients; and not reported and 92 % (72/78) in immune competent patients, respectively [46]. Overall, the sensitivity across all reviewed patient populations for serum (94 %) and CSF (94 %) antigen titers was much better than that of India ink (60 %) and similar (or slightly better) to that of CSF culture (92 %). The argument follows that for HIV-positive patients unable to undergo lumbar puncture (LP) or with vague central nervous system (CNS) symptoms not warranting an LP, a serum antigen test may be a reasonable surrogate for meningitis screening [46, 47]. On the other hand, in immune competent patients with meningitis symptoms, the utility of a serum antigen is less clear (i.e., CSF antigen is required) [48, 49].

The prognostic value of antigen titers is controversial given the above mentioned issues of disease and host factors. In the majority of studies, higher initial titers are associated with increased mortality and relapsed disease; however, other recent studies have not confirmed this association [39, 47, 50–54]. Also controversial is the use of sequential antigen titers obtained during therapy. This practice is best studied in the HIV patient population, in which setting the use of titers to monitor treatment response is probably unreliable and therefore not recommended [45, 55]. In certain situations, sites other than serum and CSF may be useful in detection of cryptococcal disease including the pleural fluid and urine [41, 56].

All cryptococcal antigen assays appear to have excellent accuracy for C. neoformans. However, because of different binding affinity to different GXM serotypes, the tests may have differing abilities to detect C. gattii (which has Serotype C GXM). In vitro data and a recent clinical report indicate that the Meridian EIA and CALAS kits may be less sensitive than IMMY LFA and EIA for C. gattii detection [40].

The only limitations of these assays are the occasional false-negative results in patients with extremely low or high Cryptococcus organism burden and infrequent false-positive results, generally resulting in low titers, in patients with other infections including disseminated trichosporonosis, Capnocytophaga canimorsus sepsis, and Stomatococcus infection [57–59].

Cryptococcus species, like Aspergillus species, also produce large amounts of mannitol, but it has not proven useful as a diagnostic marker for this disease either.

Cryptococcal infection is the rare condition where a serodiagnostic test has extremely high accuracy. A positive cryptococcal antigen result is highly suggestive of infection and can be the sole basis for initiating targeted therapy. However, definitive proof of disease still requires culture or histopathology and efforts to prove the diagnosis by these means are always warranted. The clinical utility of the antigen test depends on the extent of disease and host immune status (see Box 5.3 for recommendations). Care must be taken not to compare titers derived from different kits given the lack of standardization among manufacturers.