Determination of serum bactericidal activity against Escherichia coli by an automated photometric method

The resistance of gram-negative bacteria to complement-mediated serum activity is supposedly an important virulence factor. However, the lack of standardization in the methods used to determine serum activity and the many definitions applied make the comparisons between studies very difficult. We developed a rapid photometric method that we compared with a classical killing one. Escherichia coli in the exponential phase of growth in brain heart infusion broth (final inoculum, 10(7) CFU/ml) at 35 degrees C was added to 50% human serum in Veronal buffer. Viable counts and automatic recording of the variations in the optical densities were obtained for 40 E. coli strains isolated from the stools of healthy adults. With the viable count method, 17 (42.5%) were susceptible (at least a 1 log CFU/ml decrease), 17 (42.5%) were resistant (a 0.6 log CFU/ml increase), 4 (10%) were intermediate (poorly growing inoculum or a decrease of less than 1 log CFU/ml), and 2 could not be classified (nonreproducible results). Agreement between both methods was observed for 87.5% of the stool strains. Eight reference strains of known susceptibilities were classified identically by both methods, leading to a final concordance rate of 89.6%. A total of 129 blood isolates were tested by the photometric method: 64 (49.6%) were resistant, 50 (38.8%) were susceptible 5 (3.9%) showed early regrowth, and 10 (7.7%) were not perfectly reproducible. Of these 129 blood isolates, 5 were also tested by the killing method: 37 (49%) were resistant, 32 (43%) were susceptible, and 6 (8%) were intermediate. The concordance rate between both assays was 89% for the blood isolates; when the minor discordances were ruled out, it was 97%. This automated method could be a useful screening tool for detecting resistance to serum in clinical trials and for studying the in vitro variations of this property.