Reply to "Mistaken identity of Brucella infection".

The letter of Yang et al. (1) is an interesting account of the continuing issues associated with the misidentification of Brucella species by the current commercial identification systems. As mentioned in the letter, we experienced similar issues with the commercial bacterial identification assays that rely on phenotypic laboratory tests (2). These systems remain unreliable for identification of Brucella species. We were able to make an accurate diagnosis with the help of 16S rRNA gene sequencing in early November 2009. At that time, the database used (SmartGene) returned 100% matches with several different Brucella species, including 25 different strains. It did not detect any homology to Ochrobactrum anthropi. There are several reasons why this may have occurred. First, the specific Brucella species that was isolated in our case may have been genetically distant from Ochrobactrum anthropi and, lacking significant homology, returned no records. Second, genetic databases are continually updated as new data are acquired. Thus, over the past 3 years, the database is likely to have become more robust and capable of identifying Brucella species with homology to Ochrobactrum anthropi. Lastly, as GenBank is a public repository of sequence data, curation of said data is largely in the hands of the contributors, and uploaded sequence data can be confounded by misidentified or misnamed strains (i.e., a Brucella sequence erroneously uploaded as Ochrobactrum). Regardless of these limitations, it continues to be the strong opinion of Yang et al. that Brucella species cannot be reliably detected by the current commercial phenotypic kits on the market and that they should not be the first line of testing. In our case, this experience led us to change how we approach working with and identifying small Gram-negative rods. Since Brucella species are highly contagious, any work on small Gramnegative rods is performed in a biosafety cabinet within a negative-pressure laboratory (in our case, the Tuberculosis Laboratory). Since both Brucella species and Ochrobactrum species are positive for oxidase and catalase reactions, biochemical identification is difficult. However, not all Ochrobactrum species are positive when inoculated onto a urea slant, while all Brucella species are rapidly urea positive. This simple test can easily be performed inside a biosafety cabinet to help rule out Brucella. Our final advice is that any organism positive for oxidase and catalase that is also found to be rapidly urea positive should be handled as if it were Brucella species and identified by molecular methods that clearly distinguish between these two organisms.