Abstract
Molecular strain typing is essential for deciphering the epidemiology of Campylobacter jejuni infections. We applied two different methods, multilocus sequence typing (MLST) and analysis of the flaA short variable repeat (SVR), to 289 isolates (163 human, 56 chicken, 34 raw milk, and 36 environmental water isolates) collected in the province of Québec, Canada, over 3 years; in addition, the analysis included the pulsed-field gel electrophoresis (PFGE) typing results for a subset of 131 isolates studied previously. MLST defined 96 sequence types (STs) and 20 clonal complexes (CCs), including 49 STs (73 isolates, 25%) and 39 alleles not previously documented in an international database. The frequency of new STs was significantly higher among water isolates than among isolates from other sources (18/36 [50%] and 55/253 [22%], respectively; P < 0.001). Nine of the 10 most prevalent CCs included isolates from humans and at least one other source; five CCs comprised exclusively or mostly human and chicken isolates. However, water and milk were the predominant nonhuman sources among the remaining CCs, suggesting that sporadic C. jejuni infections in humans may frequently arise from sources other than chickens. All three typing systems were discriminatory (discriminatory index > 0.9). Among 131 isolates analyzed by PFGE, each of the 20 types represented by two or more isolates corresponded to a single CC. In contrast, among the 14 most prevalent types detected by analysis of the flaA SVR (5 to 27 isolates each), 8 (57%) included isolates that represented multiple different CCs. The basis for these discordant results was uncertain. Antimicrobial resistance was randomly distributed among the CCs and appeared to be more closely related to the source of an isolate than its genotype. Although MLST is labor-intensive and expensive, it remains the single best method for the genotyping of C. jejuni isolates and deciphering the epidemiologic relationships among isolates.