Use of multivariate analysis to compare antimicrobial agents on the basis of in vitro activity data

Multivariate techniques such as principal component analysis or similar factor analysis help in analyses of the simultaneous interrelationships among several variables. A comparative multivariate analysis on the in vitro activities of eight antimicrobial agents, including the novel molecule daptomycin, is presented. Multivariate analysis detects components or factors and establishes connections among antimicrobial agents on the basis of their different levels of participation in each factor. The first principal component was dominated by vancomycin, teicoplanin, and rifampin (0.94344, 0.92792, and 0.72127, respectively). The second principal component showed strong effects from imipenem, gentamicin, and cephalothin (0.87922, 0.86126, and 0.68870, respectively). Daptomycin stood out alone in the third principal component (0.83983). The first three components defined 81.5% of the total variance and could easily be represented graphically in a three-dimensional scatter plot. In this graphic representation, the eight antimicrobial agents clustered in three different spatial regions; daptomycin occupied a separate spatial position. The use of multivariate analysis offers a different approach to determination of the in vitro activities of new antimicrobial agents and adds some new data on the relationships among different classes. Notwithstanding its limitations, the application of these methods in microbiology and drug development could be an additional tool for use in processing information.