1. Introduction

Aspergillus species have been known for decades as important pathogens of birds, leading to high mortality rates, including the Spheniscidae family, being about 90-95% of the cases of aspergillosis caused by Aspergillus section Fumigati ( ^{Cray et al., 2009b} ; ^{Cabana, 2013} ; ^{Xavier et al., 2011} ).

The diagnosis ante-mortem of invasive aspergillosis (IA) in birds is limited and traditional techniques, such as blood tests, biochemical tests and imaging studies may reveal only nonspecific changes ( ^{Xavier et al., 2011} ; ²⁰⁰⁸ ; ²⁰⁰⁷ ).

Mycological classic tests have low sensitivity and / or specificity, and use of serological tests, although indicated in the literature ( ^{Cabana et al., 2015} ; ^{Tell, 2005} ) it is not in routine use, as the gold standard is still restricted to histopathological and mycological post-mortem ( ^{Cray et al., 2009a}, ^b ).

Modern techniques for the diagnosis of aspergillosis by direct detection of antigen galactomannan (GM) in clinical samples have been increasingly used for the diagnosis of IA in humans from different species. The GM is a polysaccharide present in the fungal cell wall of the genus Aspergillus , a family of derivatives galactofuranose antigens. Their release into the bloodstream occurs during the growth of hyphae in tissue invasion process ( ^{Mennink-Kersten et al., 2004} ; ^{Nucci and Colombo, 2012} ). This molecule can be considered as an important biomarker for the determination of invasive fungal infections by Aspergillus spp. in different clinical samples to be water soluble ( ^{Maertens et al., 2007} ; ^{Xavier et al., 2011} ).

The Platelia Aspergillus EIA® (Bio-Rad USA) is a commercially available diagnostic kit which is based on sandwich ELISA for detection of galactomannan ( ^{Maertens et al., 2007} ; ^{Xavier et al., 2011} ).This test is standard for blood serum and bronchoalveolar lavage of human and neutropenic patients, and when performed serially, anticipates the diagnosis of aspergillosis within one week. Some limitations are described, and the rates of false-negative and false-positive results fluctuate around 10% ( ^{Mennink-Kersten et al., 2004} ; ^{Nucci and Colombo, 2012} ).

Although the ELISA sandwich test for GM detection is considered an important diagnostic tool for AI in humans. Studies have shown that can also contribute to the diagnosis of other species such as dogs, cattle and poultry ( ^{Arca-Ruibal et al., 2006} ; ^{Billen et al., 2009} ; ^{Cray et al., 2009a} ; ^{Franca et al., 2012} ; ^{Garcia et al., 2001} ; ^{Garcia et al., 2008} ; ^{Guillot et al., 1999} ; ^{Jones and Orosz, 2000} ; ^{Nucci and Colombo, 2012} ; ^{Xavier et al., 2011} ). However, there are protocols and indications for its use in penguins.

Due to the high incidence of aspergillosis in penguins determining high mortality rates in these animals in captivity as well as the difficulty of the definitive diagnosis of ante-mortem disease in this species, this study aim to evaluate the effective of Platelia Aspergillus EIA® test (Bio-Rad-US) the diagnosis of aspergillosis in naturally infected Magellanic penguins, determining sensitivity, specificity, and positive and negative predictive values for different cutoff points.

2. Material and Methods

They were included in the study, blood serum samples of Magellanic penguins that died of aspergillosis during the rehabilitation period in the Centro de Recuperação de Animais Marinhos of Rio Grande - CRAM-FURG (n=29). The samples are stored in the mycology laboratory- FAMED-FURG, and all cases were confirmed from post-mortem examinations with mycological culture and histopathological examination. As a control group, were included over 23 serum samples from healthy Penguins CRAM-FURG, which they were rehabilitated and released to their natural habitat. All samples were aliquoted in biosafety cabinet to prevent contamination by airborne conidia and found themselves stored at -20 ° C.

A single blood sample each animal was included and collected from venipuncture of the cephalic vein. All samples were obtained in a maximum period of 60 days (7-69) before death (case group) and / or release (control).

The GM detection was performed on all serum samples animals included in the study according to the manufacturer's instructions. In brief, 300ul of sample was added to 100ul of Platelia Aspergillus EIA® treatment solution into microtubes and subsequently arranged in the thermoblock for heat treatment for six minutes at 120° C. Then the wells were centrifuged at 10,000xg for 10 minutes. In sequence the strips were filled with 50ul conjugate and the sample supernatant and then the plate was incubated for 90 minutes at 37° C. Then the wells were centrifuged at 10,000xg for 10 minutes. In sequence the strips were filled with 50ul conjugate and the sample supernatant and then the plate was incubated for 90 minutes at 37° C. The reaction was terminated by addition of 1.5 N solution of sulfuric acid and reading the optical density (OD) at 450 nm with 620 nm reference filter. The tests were conducted in duplicate and reactions were all positive control samples used, and the negative cutoff point provided by the kits diagnostic. The GM index was obtained by dividing the average value of OD of the duplicate of the clinical sample by duplicate OD of the average value of the cut-off sample provided by the kit.

From the database Laboratório de Micologia, It was obtained information about the presence of anti-Aspergillus fumigatus antibodies detected by agar gel immunodiffusion (AGID) for all serum samples.

Results were analyzed using chi-square test and Kruskal-Wallis from SPSS 20.0, IBM®. ROC curve was obtained and from this, rates of sensitivity, specificity, positive and negative predictive values were also calculated based on four different cutoff points (0.5, 1.0, 1.5 e 2.0).

3. Results

The serum GM index did not differ between animals in the case group and control (p _KW = 0.097). The penguins with aspergillosis (n = 29) GM index ranged from 0.29 to 6.5, with a median of 0.91 and mean of 1.71, while the healthy animals (n = 23) the median was 0.75 and 1.13 average (varying from 0.31 to 4.83) ( Figure 1 ).

Figure 1 Serum GM indices in healthy penguins (0) and aspergillosis (1). 

From these values it was determined the ROC (Receiver Operating Characteristic) for serum GM detection in the diagnosis of aspergillosis in penguins, where the x-axis (x) is the true positive (sensitivity) and the ordinate axis (y) is false positive (1- specificity) ( Figure 2 ). The analysis of the test demonstrated a precision index value of the area under the curve of 0.635 (0.482 to 0.788 CI) (p = 0.097).

Figure 2 ROC curve of serum GM detection to diagnosis of aspergillosis in penguins.  

From the values determined by the coordinate of the curve, four different cut points (0.5, 1.0, 1.5 and 2.0) were analyzed, resulting in sensitivity rates ranging from 86.2% to 34.5% and specificity between 91 3% and 26.1% ( Table 1 ).

Of the 52 animals studied, only four had positive IGA, all of them belonging to the group case. Comparing the serum GM index in group case as the presence or absence of antibodies detected by AGID was no significant difference (p = 0.503), mean and median values of 2.59 and 1.76 (± 2.80), respectively, in animals with positive AGID, and 1.36 and 0.89 (± 1.20), respectively, in animals with antibodies to A. fumigatus.

4. Discussion

This study evaluated for the first time the applicability of the commercial Aspergillus EIA® Platelia kit for the diagnosis of aspergillosis in Magellanic penguins naturally infected, finding no significant difference in GM ratios between animals with and without the disease. A single study the this population with a similar was described by ^{Cray et al. (2009b)} , however the authors included 56 birds with aspergillosis, of which only three were penguins, it is not possible to extrapolate the test efficacy results obtained by the authors for Sphenisciformes family.

In domestic and wild birds this differents families and orders, including some species of raptors, Psittaciformes, Anseriformes and Galliniformes. Described studies regarding the effectiveness of GM detection to diagnosis of aspergillosis, and demonstrate sensitivity rates ranging from 12 to 67% and specificity ranging 73 and 95% using 0.5 as the cutoff point ( ^{Arca-Ruibal et al., 2006} ; ^{Cray et al., 2009a}, ^b ; ^{Dhama et al., 2013} ; ^{Franca et al., 2012} ; ^{Fischer et al., 2014} ).

These results do not match those found in our study with penguins, when considering this same cutoff value was detected high sensitivity rate (86.2%) but low specificity (26.1%). Similar rates of the authors mentioned above were found in our study only using a cutoff point four times (2.0), in this case the sensitivity was 34.5% and specificity of 91.3%.

Factors such as different animals species included in the studies, duration and development of aspergillosis in these birds, as well as clinical presentation of the disease (infection site) and immune response of different species of birds ( ^{Deem, 2003} ; ^{Tell, 2005} ) may be related to conflicting results found in our study compared to others in the literature.

The high rate of false-positive results found in our study may be related to colonization of the respiratory tract by Aspergillus species Fumigati section, exposure to environmental strains or mainly to the presence of a cross-reactive antigen has not been elucidated, which is also suggested by other authors ( ^{Le Loch et al., 2005} ). On the other hand, false negative rates in the Platelia Aspergillus EIA® are generally related to encapsulation of the infection, immunocomplex formation by anti- Aspergillus antibodies or prior exposure to antifungal agents (for prophylaxis) ( ^{Xavier et al., 2011} ). However, none of these hypotheses can be extrapolated to our study, whereas all penguins with aspergillosis included had lesions spread through the respiratory tract to the post-mortem examination, not characterizing frame encapsulation of the infection and were not in antifungal treatment, moreover, no significant difference in the GM results comparing animals with and without antibodies to Aspergillus sp. (PKW = 0.449). However, the interference factors in the test for this animal species (penguins) are not well been elucidated, as well as other animals ( ^{Arca-Ruibal et al., 2006} ; ^{Fischer et al., 2014} ).

In attempt to reduce the false-positive rate, authors recommend testing in at least two serum collections ( ^{Arca-Ruibal et al., 2006} ; ^{Cray et al., 2009a}, ^b ; ^{Verweij et al., 1995} ). In our study, only one clinical sample per animal was included, being this a limitation, in that it was not possible to evaluate the test results when performed as serial monitoring of serum levels of GM penguins.

Our results show that serum GM detection by Platelia Aspergillus EIA® does not seem to be useful for the diagnosis of aspergillosis in naturally infected penguins, with high rates of false-positive results with cut-off 0.5 (indicated by the manufacturer) and false negatives in high cut-off.