Serovar Determination and Molecular Taxonomic Correlation in Mycobacterium avium, Mycobacterium intracellulare, and Mycobacterium scrofulaceum: A Cooperative Study of the International Working Group on Mycobacterial Taxonomy

Abstract
A cooperative study was conducted by the International Working Group on Mycobacterial Taxonomy to correlate the agglutination serovar designations of Mycobacterium avium, Mycobacterium intracellulare, and Mycobacterium scrofulaceum strains with the species ascriptions of these organisms according to molecular criteria and cultural properties and to assess the reproducibility of serovar determinations for a set of 63 reference strains of these species. Among the molecular criteria, the level of agreement between results obtained with nucleic acid probes and T-catalase serology results was 94% for strains of M. avium and M. intracellulare. Nucleic acid probes were not available for M. scrofulaceum, but none of the 10 strains ascribed to this species on the basis of catalase serology data reacted with a nucleic acid probe for M. avium or M. intracellulare. Ascription to a species on the basis of mycolic acid high-performance liquid chromatography patterns was in agreement with catalase serology results in 86% of the cases examined. Most strains belonging to serovars 1 through 6 and 8 through 11 were identified by molecular criteria as M. avium, most strains belonging to serovars 7, 12 through 20, 23, and 25 were identified as M. intracellulare, and most strains belonging to serovars 41 through 43 were identified as M. scrofulaceum, in agreement with common current practice. Evidence for assigning serovar 27 to M. scrofulaceum was obtained. However, two strains of a given serovar may, on occasion, be placed in different species. The dominant species assignments for strains belonging to serovars 21, 24, 26, and 28 remain unresolved. Data from laboratories which used panels of sera that corresponded to all of the serovars represented in the study were in agreement with the consensus results in 84% of the instances. Laboratories in which a limited panel of sera was used were far more likely to report results that did not agree with the consensus results.