Rapid Capsular Antigen Immunoassay for Diagnosis of Inhalational Anthrax: Preclinical Studies and Evaluation in a Nonhuman Primate Model

ABSTRACT Inhalational anthrax is a fatal infectious disease. Rapid and effective treatment is critically dependent on early and accurate diagnosis. Blood culture followed by identification and confirmation may take days to provide clinically relevant information. In contrast, immunoassay for a shed antigen, the capsular polypeptide gamma-d-polyglutamate (γDPGA), can provide rapid results at the point of care. In this study, a lateral flow immunoassay for γDPGA was evaluated in a robust nonhuman primate model of inhalational anthrax. The results showed that the time to a positive result with the rapid test using either serum or blood as a clinical specimen was similar to the time after infection when a blood culture became positive. In vitro testing showed that the test was equally sensitive with cultures of the three major clades of Bacillus anthracis. Cultures from other Bacillus spp. that are known to produce γDPGA also produced positive results. The test was negative with human sera from 200 normal subjects and 45 subjects with culture-confirmed nonanthrax bacterial or fungal sepsis. Taken together, the results showed that immunoassay for γDPGA is an effective surrogate for blood culture in a relevant cynomolgus monkey model of inhalational anthrax. The test would be a valuable aid in early diagnosis of anthrax, which is critical for rapid intervention and a positive outcome. Use of the test could facilitate triage of patients with signs and symptoms of anthrax in a mass-exposure incident and in low-resource settings where laboratory resources are not readily available.

can be effectively deployed, dispensed, and administered is critical to a successful response. However, a mass-casualty incident has the potential to rapidly overwhelm a local health care system and deplete essential medical countermeasures such as intravenous antibiotics and antitoxins if not effectively and selectively utilized (3,4). The availability of a rapid diagnostic test that could differentiate between active inhalational anthrax infection on the one hand and the worried well, other forms of anthrax, or individuals who are not infected would focus available resources on those individuals most in need of treatment. An inexpensive rapid diagnostic test would also facilitate diagnosis in low-resource settings where anthrax may naturally occur.
The current gold standard for diagnosis of inhalational anthrax is the blood culture, which requires 24 h to a negative culture and 3 to 4 days to confirm positive cases (5). Blood culture has high sensitivity and specificity, but the time to acquire results from culture may not be conducive to initiate early treatment. Alternatives to blood culture include immunoassay for protective antigen (PA), assay for lethal factor (LF), and PCR for the pagA gene (6)(7)(8)(9). Such tests would be used only if inhalational anthrax was suspected. Unfortunately, these tests vary in sensitivity, may require extensive sample processing, may take several hours to produce a result, may require complex instrumentation, or may require specific laboratory skills. A rapid, sensitive, and inexpensive point-of-care (POC) assay for presumptive diagnosis of inhalational anthrax complemented with clinical symptoms and epidemiological information would identify individuals most likely to benefit from early treatment that might not be possible with the currently existing tests.
Capsular antigens have been recognized as biomarkers for fungal and bacterial infections since the seminal report by Dochez and Avery that the capsular polysaccharide of Streptococcus pneumoniae can be detected via immunoassays using serum and urine from infected humans and rabbits (10). The presence of a secreted biomarker for anthrax was suggested by early studies by Ascoli and Valenti and Bail who found soluble antigen in body fluids of infected animals (cited in reference 10). B. anthracis resembles other encapsulated bacteria and fungi in the production of an antiphagocytic capsule; however, unlike other encapsulated microbes that produce polysaccharide capsules, B. anthracis produces a capsular polypeptide that is a homopolymer of D-glutamic acid residues that are linked via the g-carboxyl (gDPGA) (11).
In previous studies, we demonstrated that detection of gDPGA in blood is an early indicator of infection in murine and rabbit models of inhalational anthrax (12,13). gDPGA could be detected in serum at approximately the same time after challenge that a blood specimen subjected to culture (which required 24 h of incubation) became positive (13). These studies relied primarily on the enzyme-linked immunosorbent assay (ELISA) immunoassay platform and included use of lateral flow immunoassay (LFIA) for assessment of gDPGA in serum. The goal of the present study was to evaluate the extent to which a rapid, point-of-care immunoassay in LFIA format (Active Anthrax Detect [AAD]) could detect early-stage inhalational anthrax in a large-scale, robust, and physiologically relevant nonhuman primate (NHP) model of inhalational anthrax (14). In addition, we performed selected preclinical studies that assessed matrix effects, limits of detection, reactivity across clades of B. anthracis, reactivity with near neighbors of B. anthracis, reactivity with select bacteria which carry genes needed for synthesis of gDPGA, and reactivity with sera from healthy subjects and subjects with culture-confirmed sepsis due to nonanthrax bacteria and fungi.

RESULTS
Kinetics after challenge for (i) appearance of bacteremia, (ii) appearance of antigenemia, and (iii) alterations in hematology and blood chemistries. Ideally, diagnosis of anthrax in a bioterrorism exposure would utilize readily available clinical specimens such as whole blood or serum. As a consequence, an initial experiment assessed sensitivity of the AAD LFIA for detection of purified gDPGA in either serum or whole blood. gDPGA was spiked into and serially diluted in pooled human serum or whole blood, and reactivity was determined by visual inspection of the LFIA in strip format. The results ( Fig. 1) showed readily visible results with gDPGA concentrations as low as 0.5 ng/mL.
With evidence that the LFIA can perform well using either serum or whole blood ( Fig. 1), experiments were done to assess the kinetics for appearance of bacteremia and antigenemia after challenge, as well as alterations in several measures of hematology and clinical chemistry in a cynomolgus monkey model of inhalational anthrax. Five groups of 14 to 15 NHPs each were challenged by the pulmonary route with B. anthracis spores. Blood was collected at various times for assessment of (i) bacteremia via simple blood culture; (ii) numbers of bacteria (CFU per milliliter) via quantitative culture; (iii) presence of gDPGA in blood and serum; (iv) hematology for blood cell count, neutrophil counts, and lymphocyte counts; and (v) clinical chemistry to monitor aspartate aminotransferase, lactate dehydrogenase, and C-reactive protein. Presence of gDPGA was determined with the AAD LFIA using visual inspection as the readout. Evaluation of whole blood was done in real time at the point of sample collection. Evaluation of serum was done after separation of serum from blood. Blood was collected at 24 h after challenge and at 12-h intervals thereafter in experiments 1 and 2 and at 6-h intervals in experiments 3, 4, and 5. Time to positivity was coded as 0 (24 h), 1 (30 or 36 h), 2 (42 or 48 h), or 3 (.48 h).
Results from our testing showed that the blood culture and the AAD LFIA for gDPGA in blood and serum became positive early in the course of infection, approximately 24 to 36 h after challenge ( Fig. 2A). There was no significant difference between the time of the first positive blood culture and a positive test for circulating gDPGA (P . 0.10 for serum or blood gDPGA versus blood culture). These results indicate that testing of serum or blood with the AAD LFIA is a rapid and effective surrogate for the blood culture. However, the AAD LFIA was able to provide a result in approximately 15 to 30 min after sample collection versus 24 h for blood culture.
Previous studies of the NHP model showed that a number of hematological and clinical chemistry parameters became abnormal in the course of inhalational anthrax (14). In agreement with this previous study, results from the present study found abnormal levels of aspartate aminotransferase, lactate dehydrogenase, C-reactive protein, white blood cell count, neutrophil counts, and lymphocyte counts late in the course of infection. All such abnormal values occurred 2 to 3 days after challenge, Rapid Diagnosis of Inhalational Anthrax in an NHP Model mBio significantly later than either the first positive blood culture or the presence of detectable gDPGA in blood and serum (P , 0.0001) ( Fig. 2A).
Quantitative blood culture was also done to determine the level of bacteremia (CFU per milliliter) at the time of the first positive test for antigenemia using serum or whole blood as a specimen (Fig. 2B). The median concentration of bacteria in blood at the time of the first positive test for gDPGA in whole blood was 3,370 CFU/mL (first quartile [Q1] = 940; Q3 = 7,130). The median concentration of bacteria at the time of the first positive test for gDPGA in serum was 1,330 CFU/mL (Q1 = 250; Q3 = 4,070). The difference was significant (P , 0.0001).
Impact of sample matrix and immunoassay format on assay performance. The AAD LFIA was developed in both dipstick and cassette formats. Both have advantages and disadvantages. As a consequence, the limit of detection for both formats was assessed using purified gDPGA that had been added to a variety of specimen matrices that might be considered for clinical use. Specimen matrices included serum; whole blood that had been treated with heparin, EDTA, or citrate; and capillary blood that had been treated with heparin or EDTA. Stock serial 2-fold dilutions of gDPGA in phosphate-buffered saline (PBS) prepared at 10Â final desired concentration were used to spike each matrix; final gDPGA concentrations in each matrix ranged from 2 ng/mL to 0.135 ng/mL. Replicates were evaluated with each LFIA format at each antigen concentration, and the results were read by two independent blinded readers. The results showed a limit of detection of 0.5 to 1.0 ng/mL regardless of the assay format or specimen matrix (Table 1). Detailed results are provided in Tables S1 and S2 in the supplemental material.
Reactivity of the AAD LFIA across clades of B. anthracis and with near neighbors. Cultures were prepared from strains of B. anthracis from each of the three major clades of the bacterium (15). Serial dilutions were prepared using human serum as a matrix, and the reactivity was determined using the AAD LFIA. The results showed a high sensitivity for detection of B. anthracis from all three clades at bacterial concentrations as low as 100 to 1,000 CFU/mL ( Table 2). Further studies were done to determine whether near neighbors of B. anthracis and other Bacillus spp. that do or do not produce  The ability of cultures to produce a positive result in an immunoassay for gDPGA could be due to gDPGA that is either bound to the bacterial cells or shed or secreted, or both. Consequently, an experiment was done to assess the relative contributions of soluble versus cell-bound gDPGA to reactivity of bacterial cultures. Cultures of B. anthracis Ames were prepared, and serial dilutions were assessed using either filtered or unfiltered cells in a human serum matrix. The AAD LFIA results (Table 4) were the same for filtered and unfiltered cultures, indicating that shed gDPGA alone can account for the positive results.
Failure to detect PGA with non-Bacillus bacteria with genes that code for cDPGA synthesis. A bioinformatics analysis by Kocianova et al. found that Staphylococcus epidermidis, which contains a polyglutamate capsule consisting of approximately equal amounts of D-and L-glutamic acid, in contrast to B. anthracis, which contains only D-glutamic acid, contains the genes that code for gDPGA synthesis (17). Homologs of the cap locus are also found in a limited number of human patho-   (Table S3). Clinical specificity. AAD in strip format was evaluated using sera from 200 normal subjects that were purchased from a commercial source. All tests were negative (data not shown). In addition, serum was obtained via a commercial source from 45 patients with a diagnosis of sepsis produced by 19 different bacteria or fungi that included Bacillus spp. (not anthracis), Candida tropicalis, Escherichia coli, Staphylococcus aureus, S. epidermidis, and Streptococcus pneumoniae (full list in Table S4). All tested negative in the AAD LFIA.

DISCUSSION
Our results indicate that testing for gDPGA in blood is an effective surrogate for blood culture for early diagnosis of inhalational anthrax in a relevant animal model. The time to a positive result using either serum or whole blood as a clinical specimen was similar to the time after infection when a blood culture became positive (P . 0.10 for serum or whole blood versus culture). Moreover, the concentration of bacteria in blood at the time of the first positive AAD result (1,300 to 3,400 CFU/mL) (Fig. 2) was similar to the concentration of bacteria in culture that produced a positive result (approximately 1,000 CFU/mL) ( Table 2). However, in contrast to the blood culture, which can take days to a confirmed positive result, the AAD test provided results in 15 min without the need for a sophisticated microbiology laboratory. Importantly, AAD provided rapid results when whole blood was collected from animals and the test was performed directly at the site of collection with appropriate proper personal protective equipment and a medical/hazardous waste management system.
The AAD test meets the ASSURED criteria that describe a diagnostic that is well suited for use in both developed and resource-limited environments (18). Lateral flow tests are typically low-complexity tests and highly affordable (18). Sensitivity was similar to that of the blood culture in the NHP model (Fig. 1). Although further testing is needed, specificity testing showed no reactivity with normal sera or sera from patients with sepsis caused by a variety of bacteria (see Table S4 in the supplemental material).  AAD is user friendly, requiring few steps, and has a platform technology that is familiar to most potential users. Results are rapid (15 min) and robust, perform well with serum or whole blood, and are compatible with available anticoagulants. In addition, identical results were observed when the test was performed in a strip format and in a strip that was enclosed in a cassette. The test requires no equipment and is read visually.
Deliverability is yet to be determined but would be similar to that of other rapid tests.
Although inhalational anthrax is one of the most severe of the biothreats, the actual frequency of inhalational anthrax in the absence of a deliberate release is quite rare. As a consequence, studies of the pathophysiology, diagnosis, and treatment of inhalational anthrax have had to rely on animal model studies. The FDA Animal Rule was developed for evaluation of new drugs when human efficacy studies are not ethical and field trials after an accidental or deliberate exposure are not feasible (FDA Animal Rule; 21 CFR 314.600 for drugs and 21 CFR 601.90 for biological products). The cynomolgus macaque anthrax inhalational model used in this study is well described and has been proposed for testing potential anthrax therapeutics in accordance with the FDA Animal Rule (14). The present study is a robust, large-scale (73 animals) application of this model to evaluate the AAD diagnostic LFIA.
We previously reported that gDPGA in serum is a biomarker for anthrax infection in murine, rabbit, and nonhuman primate models of inhalational anthrax (12,13,19). These earlier studies used serum or plasma as a specimen and an ELISA for detection of gDPGA. In the present study, we demonstrate that a point-of-care lateral flow assay can be used in real time for diagnosis of early-stage infection. The test can be performed with whole blood at the location where blood is collected without the need for a laboratory. We also provide results from in vitro studies for specificity, inclusivity, and sensitivity (CFU per milliliter). Finally, the present study used a large number of animals to provide confidence in our conclusions.
Inclusivity studies done in vitro found similar positive AAD results with representative strains of the three major clades of B. anthracis. Positive results were also found with cultures of other Bacillus spp. that are known to produce capsules demonstrable by immunofluorescence, including B. cereus biovar anthracis, B. megaterium, and B. frigoritolerans. Positive results with isolates of B. cereus biovar anthracis from Côte d'Ivoire (CI) and Cameroon (CA) are important because B. cereus biovar anthracis produces an anthrax-like disease in wild and domestic mammals in tropical Africa and is a Tier 1 select agent. No reactivity was observed with a culture of B. cereus which does not produce an observable capsule in vitro, and very weak reactivity was found with B. thuringiensis. B. thuringiensis is an insect pathogen that typically does not produce a gDPGA capsule; however, a gDPGA capsule has been noted in at least one strain (20).
One potential concern with detection of gDPGA as a surrogate for the blood culture is the possibility of false-positive reactions. As noted above, production of gDPGA is not limited to B. anthracis. For example, Bacillus licheniformis produces a capsular polypeptide that is comprised of gPGA in the D-and L-isoforms (21,22). In addition, some bacteria contain genes that code for gDPGA synthesis (17). However, available evidence suggests that the potential for false-positive results from non-anthrax-producing bacteria may be minimal. First, most non-anthracis species of Bacillus such as B. megaterium and B. frigoritolerans do not cause clinical disease that might be confused as anthrax. Pseudoinfection due to contamination of specimens by Bacillus spp. is far more likely than a true infection (23). Second, positive results were not observed with cultures of S. epidermidis or Leptospira spp. (Table S3), bacteria that have genes that code for gDPGA production (17). Finally, no false-positive results were found with sera from 200 healthy subjects and 45 patients with a diagnosis of microbial sepsis that included S. epidermidis and Bacillus spp. (not anthracis).
There is a caveat to use of blood culture for diagnosis of the earliest phases of bacteremia in the NHP model. Serial sampling of blood in the NHP model is limited to relatively small blood volumes at each time point. In contrast, Clinical and Laboratory Standards Institute (CLSI) guidelines for blood culture for human infection recommend collection of 4 to 6 10-mL bottles for evaluation of bacteremia (24). As a consequence, blood culture in the NHP model may miss very low levels of bacteremia compared to culture using CLSI guidelines due to differences in sample volumes. However, blood culture takes as many as 24 h for positive growth and 3 to 4 days to confirm identification of B. anthracis and is not suited for rapid POC use. This difference in specimen volumes between the NHP model and CLSI guidelines suggests that testing for gDPGA may not be well suited for screening of asymptomatic individuals in the earliest stages of infection. However, the AAD test would likely be a rapid and effective surrogate for blood culture in the symptomatic patient.
There are multiple applications of the AAD test for diagnosis of anthrax. First, the test could be used in a mass-exposure incident for triage of patients with signs and symptoms of inhalational anthrax. Second, the test could identify patients who would best benefit from use of countermeasures that are in short supply. Third, the test could be of value in cases of inhalational anthrax where blood cultures are negative due to receiving prior antibiotics. Fourth, a test that meets the ASSURED criteria for a diagnostic could be of great value for diagnosis of anthrax in resource-limited settings. Finally, anthrax is enzootic in many regions (25). Humans may become infected by exposure to or through consumption of meat from infected animals. AAD could rapidly identify infected animal carcasses (26).
In summary, AAD is a rapid, sensitive, and user-friendly tool for diagnosis of anthrax. The test works by detection of capsular antigen that is shed into blood during infection. Unlike the blood culture, the lateral flow immunoassay can provide results in 15 min at the point of sample collection without the need for a microbiology laboratory. The test has the potential to rapidly triage patients who have been exposed to the microbe and have symptoms suggestive of anthrax. Such use would enhance patient care and identify patients most in need of medical countermeasures such as toxin-neutralizing antibodies that might be in short supply. Finally, gDPGA is an attractive biomarker for infection because the gDPGA is an essential virulence factor that cannot be engineered to avoid surveillance (27). NHP model of inhalational anthrax. All animal procedures were approved by Battelle's Institutional Animal Care and Use Committee, and all exposures and assays were performed in a biosafety level 3 laboratory registered with the Centers for Disease Control and Prevention.

MATERIALS AND METHODS
Cynomolgus macaques (Macaca fascicularis) were aerosol challenged with a targeted dose of 200 50% lethal doses (LD 50 ) of B. anthracis Ames strain spores (1.24 Â 10 7 spores total [28]). Details regarding the inhalational model have been described previously (14). Venous blood was collected beginning 24 h postchallenge from NHPs included in an NIAID-funded study to evaluate potential medical countermeasures against anthrax. Residual blood samples that were analyzed by LFIA represented a subset of the animals in the original study. The animals analyzed by LFIA were divided into five groups of 14 to 15 NHPs for a total of 73 animals. In experiments 1 and 2, blood was collected for analysis at 12-h intervals. In experiments 3, 4, and 5, blood was collected at 6-h intervals. Routine qualitative blood culture assay was performed by three-phase streaking of 30 to 40 mL of blood onto a blood agar plate. For quantitative bacteremia culture, undiluted blood as well as serial dilutions of 1:10 was prepared and spread plate in triplicate over Trypticase soy agar (TSA) plates (100 mL/plate). Growth of colonies with morphology consistent with B. anthracis was indicated as positive or negative for standard blood culture. Colonies present on plates prepared for quantitative bacteremia were counted and reported as CFU per milliliter. The lower limit of quantitation was considered #250 CFU/mL. Standard clinical pathology (clinical chemistry and hematology) parameters were also assessed at each time point. Parameters of interest included white blood cells, polymorphonuclear leukocytes, lymphocytes, aspartate aminotransferase, Creactive protein, and lactate dehydrogenase.
Statistics. For time-to-positivity analysis, collection times were coded as 0 = 24 h, 1 = 30 or 36 h, 2 = 42 or 48 h, and 3 = .48 h (including censored). For comparisons to culture on time to positivity, a repeatedmeasures analysis was done using a marginal linear model with unstructured covariance, with coded collection time as the dependent variable and test (g DPGA serum, g DPGA blood, etc.) as the fixed effect. Dunnett's test was used to control for multiplicity. For paired comparisons of median bacteremia level at first positive test in blood versus serum, a signed-rank test was used. Hypothesis tests were two-sided with 0.05 significance levels. Statistical analyses were done using SAS v9.4 (SAS Institute Inc., Cary, NC).
cDPGA production by cultures of B. anthracis and growth conditions of other bacteria. g DPGA was isolated from B. anthracis Pasteur as previously described (12,29) and used for spiking experiments in the different matrices. For testing the AAD LFIA, spore-forming bacteria were streaked for isolation onto Trypticase soy agar plates containing 5% sheep blood (SBA) from frozen stocks stored in 25% glycerol. Multiple single colonies were used to inoculate capsule-production medium (heart infusion broth containing horse serum and 8% sodium bicarbonate) and incubated for approximately 18 to 20 h at 37°C in 5% CO 2 . Dilutions (1:10) up to 10 7 were performed in pooled human sera and tested using the AAD LFIA. Bacterial concentrations (CFU per milliliter) were calculated by plating 100 mL of dilutions in triplicate onto SBA and incubating overnight at 37°C. Dilution plates with countable colonies (30 to 300) were averaged per dilution.
S. epidermidis strains were streaked for isolation and inoculated into capsule-producing medium as described above, except that tryptic soy broth with and without 1 M sodium chloride was used, and incubation was with shaking at 37°C for 24 h in ambient atmosphere. Leptospira spp. were grown in liquid EMJH medium at 30°C for 4 days, and turbidity was adjusted to a 0.5 McFarland standard.
Lateral flow immunoassay forcDPGA. The AAD LFIA was provided by InBios, International. AAD is an antigen-capture immunoassay that is constructed from monoclonal antibody (MAb) 8B10, which is reactive with g DPGA. Details regarding construction and use of the AAD LFIA have been described previously (13,26). Lateral flow assays were constructed in both dipstick and cassette formats. The LFIA testing of NHP samples was done solely with assays in cassette format. Assays were read visually 15 min after application of the clinical sample.
Human specimens. Serum, whole blood, and capillary blood were purchased from a commercial vendor (BioIVT, Durham, NC). Pooled normal human serum and sera from 200 individuals were also purchased from BioVT. Additionally, 45 serum samples from culture-confirmed sepsis patients were acquired from BioIVT.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.