Linear Regression Equations to Predict Β-Lactam, Macrolide, Lincosamide, and Fluoroquinolone MICs from Molecular Antimicrobial Resistance Determinants in Streptococcus Pneumoniae.

Antimicrobial resistance in Streptococcus pneumoniae represents a threat to public health, and monitoring the dissemination of resistant strains is essential to guiding health policy. Multiple-variable linear regression modeling was used to determine the contributions of molecular antimicrobial resistance determinants to antimicrobial MICs for penicillin, ceftriaxone, erythromycin, clarithromycin, clindamycin, levofloxacin, and trimethoprim-sulfamethoxazole. Training data sets consisting of Canadian S. pneumoniae isolates obtained from 1995 to 2019 were used to generate multiplevariable linear regression equations for each antimicrobial. The regression equations were then applied to validation data sets of Canadian (n = 439) and U.S. (n = 607 and n = 747) isolates. The MICs for beta-lactam antimicrobials were fully explained by amino acid substitutions in motif regions of the penicillin binding proteins PBP1a, PPB2b, and PBP2x. Accuracies of predicted MICs within 1 doubling dilution to phenotypically determined MICs were 97.4% for penicillin, 98.2% for ceftriaxone, 94.8% for erythromycin, 96.6% for clarithromycin, 98.2% for clindamycin, 100% for levofloxacin, and 98.8% for trimethoprim-sulfamethoxazole, with an overall sensitivity of 95.8% and specificity of 98.0%. Accuracies of predicted MICs to the phenotypically determined MICs were similar to those of phenotype-only MIC comparison studies. The ability to acquire detailed antimicrobial resistance information directly from molecular determinants will facilitate the transition from routine phenotypic testing to whole-genome sequencing analysis and can fill the surveillance gap in an era of increased reliance on nucleic acid assay diagnostics to better monitor the dynamics of S. pneumoniae.