Robustness of a FT-IR ANN-based technique in the identification of new species of the Burkholderia cepacia Complex

ALEJANDRA BOSCH, PABLO MARTINA, JOSÉ DEGROSSI, ALEJANDRO MIÑÁN, VIVIANA FANESSI, LAURA GALANTERNIK, PATRICIA MONTANARO, CARLOS VAY, MIRTA FRANCO, JUERGEN SCHMITT, DIETER NAUMANN AND OSVALDO YANTORNO*

Sâo Jose dos Campos, Brasil

Workshop; SPEC 2008 - Shedding Light on Disease Optical Diagnosis for the new Millennium; 2008

SPEC 2008

Members of the genus Burkholderia are versatile organisms that occupy a surprisingly wide range of ecological niches. They are routinely isolated from the natural environment, where they have diverse beneficial properties. However, all Burkholderia can be opportunistic pathogens in Humans. Indeed, bacteria of the Burkholderia cepacia Complex (BCC) are responsible for devastating lung infections in cystic fibrosis (CF) patients and various infections in inmunocompromised non-CF patients. In common with many other opportunistic pathogens, it appears that severe disease and death dependent on the clinical state and predisposition of the patients at the time of infection. BCC bacteria used to be mainly patient-to-patient transmitted. Nevertheless, due to the implementation of strict infection control practices in hospitals in the last years, this source of infection was reduced. Indeed, recent reviews, describing the current status of BCC in CF infections, has clearly demonstrated that the environment and industries represent the reservoirs of new pathogenic strains for both CF and non-CF patients . Therefore, the epidemiology of BCC has changed in the last years. The isolation and reliable identification of BCC bacteria is still a serious problem in clinical microbiology laboratories, it is very limited in environmental samples, and still not implemented in industrial settings. Moreover, the taxonomic complexity of BCC and its dramatic evolution has contributed to unreliable identifications. Even though 9 distinct species of closely related bacteria (genetic types or genomovars) are known to comprise what is referred to as the BCC, recently, genetic methods such as recA gene sequence analysis and multilocus sequence typing (MLST), have allowed the description of 6 additional novel species and one group of BCC isolates known as Group K with two novel species (B. contaminans and B. lata). The aim of present work is to show the possibilities of Artificial Neural Networks – FTIR spectroscopy-based systems to accurate and rapidly identify environmental and clinical BCC isolates at the species level. After 3 years of working in collaboration with different CF health care centers in Argentina (2004-2007), we proved the high reliability and strong potential of ANNs- FT-IR spectroscopy-based methodology for a rapid identification of gram-negative rods suitable for routine microbiology diagnosis in CF patients.  In that research recA-RFLP and species-specific-PCR methodologies were used as gold standards identification techniques. Throughout 2007 and 2008 we have applied in both clinical isolates (CF and Non-CF patients) and industrial products and settings, recA sequencing analysis. Firstly, we found that the clinical isolates previously identified as the B. cenocepacia strains with “atypical RFLP pattern”, indeed belong to the Group K, one of the new groups of bacteria mentioned above. Therefore, secondly, we updated our multivariate identification system which have showed robustness in identification of isolates recovered form both CF and non-CF patients. Finally, as we have found that the phenotypic characteristics of clinical and environmental isolates belonging to the same of BCC specie were significantly different, the ANNs-FTIR based system developed for clinical isolates was not accurate for the identification of environmental isolates. Therefore, a different multivariate identification system is being developed for the characterization and identification of environmental isolates. These results are currently serving in the microbiology diagnosis of CF patients, and they will contribute, together with other cooperative techniques, to determine therapeutic options and infection control measures in patients. Moreover, the construction of a new data base with environmental isolates spectra, and the development of a multivariate identification system will help to perform a systematic study focusing to find the main source of BCC infections.