Abstract
In cases of chronic Brucella spp. infection, results of the rose bengal plate test (RBPT) and indirect enzyme-linked immunosorbent assay (ELISA) should be coherent, as reported in controlled conditions in the literature. We compared RBPT and ELISA results in 58 Alaska grizzly bears (Ursus arctos horribilis), eight Kodiak brown bears (Ursus arctos middendorffi), and six Alaska Peninsula brown bears (Ursus arctos gyas). Of the 72 bears tested, 42 (58%) were ELISA positive and 53 (73%) were RBPT positive. However, the coherence between the tests was only fair (K=0.37, SE=0.11), suggesting that either the serologic results were not compatible with Brucella spp. infection or that there was a technical problem with the tests. To address a potential technical problem, we performed a 30-min chloroform/centrifugation cleanup. Following cleanup, the ELISA identified 43 positives (59%) and the RBPT identified 47 (65%), and the coherence between the tests was much improved (K=0.80, SE=0.07). We recommend cleaning wildlife sera with a high lipid content before performing RBPT and performing RBPT and ELISA in parallel to assess coherence. Our results suggest that Alaskan brown bears have been exposed to Brucella spp.
Classically, serologic tests for detecting antibodies against a specific etiologic agent or group of agents are the first screening tools used in humans, livestock, and wildlife. The presence of antibodies indicates exposure to an agent, which may be due to a current infection or an earlier infection. To make sound inferences based on serologic test results, there are two prerequisites: 1) the biology of the infection needs to be understood (e.g., if the host species is a reservoir or a spillover host; Godfroid et al. 2014), and 2) the test, often developed for use in a livestock species, needs to be validated when used in other species (World Organisation for Animal Health [OIE] 2015a). These prerequisites are especially important when studies rely solely on serologic results, which is often the case in wildlife disease studies (OIE 2015b). Serologic tests can be classified into species-specific (e.g., brucellosis indirect enzyme-linked immunosorbent assay [ELISA], although such tests may be used in different species with an appropriate conjugate) or species-nonspecific (e.g., agglutination tests like the rose bengal plate test [RBPT] commonly used in humans, livestock, and wildlife; Godfroid et al. 2010, 2014).
The serologic tests recommended by the OIE for detecting exposure to Brucella spp. use antigens derived from Brucella abortus (OIE 2015c). This is because the immunodominant antigens of Brucella spp. are associated with the smooth lipopolysaccharide (S-LPS) and are shared to a very large extent by the smooth Brucella species. Consequently, it is impossible to ascribe to which smooth Brucella species the antibodies detected in a host are directed (Godfroid et al. 2014).
Another problem with brucellosis serology is cross-reactivity. Some Gram-negative bacteria (the most important being Yersinia enterocolitica O:9) have an S-LPS bearing some of the same immunodominant antigens as Brucella S-LPS. In case of exposure to Y. enterocolitica O:9, using multiple brucellosis serologic tests will not solve the problem (Godfroid et al. 2014). Indeed, all validated and recommended brucellosis serologic tests use antigens derived from B. abortus biovar 1, which shares the A epitope associated to the S-LPS with Y. enterocolitica O:9 (OIE 2015c) and will thus classify the serum as positive.
Our goal herein is not to provide a test validation study but to highlight a technical pitfall when using brucellosis serologic tests in wildlife species with a high lipid content in sera. For illustration, we compare Alaska grizzly bears (GB; Ursus arctos horribilis), Kodiak brown bears (KB; Ursus arctos middendorffi), and Alaska Peninsula brown bears (APBB; Ursus arctos gyas) ELISA- and RBPT-results before and after a 30-min chloroform/centrifugation cleanup of the sera (Castro et al. 2000; Blanchet et al. 2014).
All sera were tested for anti-Brucella antibodies with an indirect ELISA validated for the detection of anti-Brucella antibodies in polar bears (Ursus maritimus; Nymo et al. 2013) and with the RBPT (IDEXX Laboratories, Pourquier, Hoofddorp, the Netherlands). The chloroform/centrifugation cleanup was as described by Castro et al. (2000) and Blanchet et al. (2014). Samples were subsequently reanalyzed with the ELISA (referred to as ELISA+) and RBPT (referred to as RBPT+) after cleanup. Statistics were done in JMP 11 (SAS Institute, Medmenham Marlow, UK). Pairwise coherence among the tests was assessed using Cohen's Kappa (K; Cohen 1960, 1968).
Blood samples (n=72) were from GBs (n=58, 22 males, 33 females, three with unknown sex), KBs (n=8, four males, four females), and APBBs (n=6, two males, four females) from Alaska game management units (Fig. 1) sampled in 1974–2008 (GB), 1982–88 (KBs), and 1986 (APBBs). The ages of bears (years) were 1–26 (GBs), 1–19 (KB), and 3–6 (APBBs).
Of the 72 total samples, 42 (58%) were ELISA-positive (36 GBs and six KBs) and 53 (74%) were RBPT-positive (41 GBs, seven KBs, and five APBBs). The coherence between the tests was only fair (K=0.37, SE=0.11). In acute Brucella spp. infections, a positive RBPT result usually appears within 15 d, followed by a positive ELISA result within 1 wk, whereas in chronic infections (>1 mo) both RBPT and ELISA are positive for prolonged periods, as demonstrated in mice and livestock (Godfroid et al. 2010, 2014; Nymo et al. 2014). Hence, in the case of Brucella spp. infection in wildlife, there should be agreement between most RBPT and ELISA results, as it is probable that the animals have experienced an infection of 1 mo or more (Godfroid et al. 2010). Our results suggest that either the serologic results are not compatible with Brucella spp. infection or that there is a technical problem with the tests. In order to address the technical problem, we performed the chloroform/centrifugation cleanup protocol.
Following cleanup, the ELISA+ identified 43 positives (60%, 37 GBs and six KBs), the RBPT+ identified 47 (65%, 40 GBs, six KBs, and one APBB), and the coherence between the tests was good (K=0.82, SE=0.07). The ELISA and the ELISA+ were in very close coherence (K=0.97, SE=0.03), showing that the chloroform clean-up had little effect on the ELISA results. The coherence between the RBPT and the RBPT+ was only moderate (K=0.55, SE=0.11) due to four samples that were negative by the RBPT and positive by the RBPT+ and 10 samples positive by RBPT and negative by RBPT+. It is unclear why four negative samples in RBPT became positive in the RBPT+. Besides lipid removal (60–62%), proteins are also removed (3.07–9.42%) by chloroform cleanup (Castro et al. 2000). Castro et al. (2000) showed that chloroform treatment had an effect on serologic tests for syphilis-detecting immunoglobulin M (IgM), with one sample changing from reactive to equivocal and another changing from equivocal to nonreactive, indicating a loss of IgM. In the context of brucellosis, nonspecific agglutination is mainly due to IgM; the low pH in the RBPT prevents some agglutination by IgM and encourages agglutination by immunoglobulin G (IgG)1, thereby reducing nonspecific interactions (Nielsen 2002). We therefore speculate that some of the nonspecific IgMs were eliminated by chloroform cleanup, allowing specific anti-Brucella IgGs to agglutinate in the RBPT, but the factors that enhance the agglutination of anti-Brucella IgGs after chloroform cleanup in the RBPT are not known.
The RBPT is a simple and reliable test recommended by the OIE. However, when using it in wildlife, two factors may interfere with the partly subjective reading of the agglutination and thus contribute to a biased result. These are the presence of hemolysis (which interferes with accurate reading) and fat globules being wrongly identified as agglutinates. Therefore, we recommend comparing RBPT to ELISA results and, if large discrepancies are found, performing a chloroform/centrifugation cleaning. Previously published results, based solely on agglutination tests (Zarnke et al. 2006; Di Francesco et al. 2015) should be carefully evaluated.
Regarding brucellosis in bears in Alaska, the RBPT+, ELISA, and ELISA+ results were coherent, suggesting that exposure to smooth Brucella spp. is likely to have occurred throughout the state except in game management unit (GMU) 9E, where no antibody-positive APBBs (n=6) were detected (Fig. 1). However, it remains to be documented which Brucella species was detected. In Alaska, Brucella suis biovar 4 has been isolated from reindeer and caribou (Rangifer spp.), and the infection might spill over to bears (Neiland 1975; Neiland and Miller 1981). The GBs in GMU 4 and KBs in GMU 8 have no contact with caribou, and only a remnant population of feral reindeer are present on Kodiak Island. However, we identified one ELISA+ positive GB in GMU 4 and six ELISA+ positive KBs in GMU 8 (Fig. 1), suggesting a possible spillover from marine mammals (Nymo et al. 2011). Bacteriology, the only certain diagnosis, should be performed in future studies to identify the Brucella species to which bears were exposed and to exclude exposure to cross-reactive bacterial species in brucellosis serologic tests.
We thank E. Breines and E. Hareide at UiT – The Arctic University of Norway, Research Group for Arctic Infection Biology, for excellent laboratory work. Thanks to H. Reynolds, J. Hechtel, J. Schoen, R. Shideler, R. Zarnke, and the many other Alaska Department of Fish and Game (ADFG) biologists who collected and archived serum samples and associated data. We are also grateful to N. Pamperin, ADFG, for graphical assistance.