The results presented herein emphasise the importance of choosing the correct identification test for accurate speciation of staphylococcal species of animal origin. The accurate identification of staphylococcal species impacts directly and positively on treatment outcomes and on the epidemiological analysis of emerging trends in multi-drug resistant staphylococcal infections in veterinary medicine.
This present study revealed that all three phenotypic test systems yielded inaccurate speciation results when compared to tuf genotyping (Table 3). When considering phenotypic test kits on their own, one must consider the reliability of reading a result with a high “apparent” accuracy. For example, RapID identified one isolate as S. xylosus with a 97% probability value; however this isolate was subsequently identified by tuf as S. pseudintermedius, which could mislead the diagnostician. One of the arguments for using phenotypic test kits is that they are less costly than genotyping. When comparing the costs of phenotyping one must consider the potential consequences of misidentification including unnecessary morbidity and mortality of infected animals.
Of the phenotypic tests utilized, the API Staph 32 correctly identified 100% of S. aureus isolates, 83.3% of SIG isolates and 31% of the CONS; the RapID test correctly identified 81.8% of S. aureus, 33% of SIG isolates and 6.8% of the CONS; while the Staph-Zym test correctly identified only 2% of all isolates. Each of these tests is based on the evaluation of expression of genetically encoded characteristics by bacterial isolates. Inaccurate speciation may be due to variable expression of biochemical traits within species, as previously reported by Blaiotta et al. . This is supported in the present study where it was observed that in each of the test systems, some biochemical tests frequently gave a misleading response for a given species tested. In particular, tests for arginine dihydrolase, arginine arlyamidase, β-glucuronidase, fructose and mannitol fermentation, novobiocin resistance and nitrate reduction, were observed in one or more systems to generate a response contrary to the expected result for a given species (data not shown). In addition, species identification kits such as these are manufactured for the human diagnostics market and are interpreted against databases with reference strains of human origin. This suggests that the reproducibility and therefore reliability of these tests is questionable when applied to veterinary isolates. When the identification of S. aureus by the three phenotyping test kits is considered, it was observed that while all of the S. aureus isolates were identified by API Staph 32, they were not consistently identified by either Rap-ID or Staph-Zym, demonstrating a lack of correlation between tests systems for a commonly isolated species. In addition, of the eight S. equorum isolates identified by genotyping, API Staph 32 identified four as S. xylosus, one as S. epidermidis, one as S. warneri and did not identify two, suggesting within species variability for the test system. With respect to the newly recognised Staphylococcus intermedius group , the failure of the phenotypic test kits to correctly speciate members of this group is of concern due to the relevance of S. pseudintermedius, not only as a veterinary pathogen, but as a source of nosocomial infection .
DNA sequencing of housekeeping genes is regularly used to definitively type staphylococcal isolates, [9, 18–21, 32]. In this study, tuf and rpoB gene segments were amplified by PCR and sequenced according to published methods ([9, 19], respectively). Both genes constitute more discriminatory targets than the 16S-rRNA gene to differentiate closely related staphylococcal species. The results of this study demonstrated 100% accuracy for reference strains using tuf genotyping. Among the clinical isolates, 23 were identified by tuf genotyping as coagulase positive species (Table 3). Interestingly, five of these isolates failed to coagulate in both the tube and slide agglutination tests. These findings are not atypical however. According to Murray et al.  up to 30% of S. aureus field isolates fail to display coagulase activity. Reduced coagulase activity in S. aureus is also reported to be associated with reduced susceptibility to vancomycin . It is worth noting that in the present study, each of the four S. pseudintermedius isolates which failed to coagulate also showed reduced susceptibility to vancomycin (data not shown). tuf genotyping identified 29 coagulase negative isolates and 5 distinct species (Table 3). The species S. carnosus and S. simulans, however, could not be definitively differentiated from each other by tuf genotyping. Likewise, S. warneri and S. pasteuri were not differentiated by this method. Previous authors have documented a close phylogenetic link between these pairs of species [29, 34, 35] and the current findings support this.
All three of the reference SIG species were identified correctly by tuf genotyping (S. intermedius, S. delphini, S. pseudintermedius). Ten of the clinical isolates were identified as S. pseudintermedius, but for two additional isolates, tuf could not differentiate between S. pseudintermedius and S. delphini. In an attempt to clarify the identity of these two isolates, both of canine origin, rpo B sequencing and PCR- RFLP of the pta gene were performed. Both isolates were confirmed as S. pseudintermedius by the two methods. Given the clinical significance of S. pseudintermedius in veterinary medicine, and the published evidence that MRSP (methicillin resistant S. pseudintermedius) is emerging as a nosocomial infection, the importance of an accurate identification is paramount. Our findings suggest the use of rpo B genotyping or PCR-RFLP of the pta gene as a confirmatory method for discriminating between SIG isolates until a larger cohort of these species are entered into the tuf gene database, thereby enhancing its accuracy.