Mycobacteria Identification By HPLC Pattern Recognition

In 1984, the United States had the lowest tuberculosis infection rate in modern history. In 1985, TB incidence started rising and has continued to rise. Additionally, TB resistance to drugs has increased, from 10 percent in the early 1980's to 23 percent in 1991 for one standard drug and from 3 percent to 7 percent for multi-drug-resistant TB. Current identification procedures utilize growth and enzymatic tests to distinguish between bacterial genera and species. These tests can add several weeks of uncertainty onto the time needed to culture colonies to testable sizes and fully identify the microbial species when tuberculosis related. This delay may be crucial for treatment regimens needed for drug resistant strains. Mycolic acids in the cell walls of mycobacteria have been shown to be indicative of individual species. High Performance Liquid Chromatography (HPLC) is a useful analytical method for obtaining distinct chromatographic patterns of these acids. Pattern recognition methods have been applied to develop 100% accurate classification models for 188 samples of 8 mycobacteria species. We propose to extend these models to identify 36 species. Preliminary indications are that this approach is as fast as genetic probes after colony growth at 1/4 of the cost, or less.