Revisiting the inoculum effect for Streptococcus pyogenes with a hollow fibre infection model.

Severe, invasive Streptococcus pyogenes (Strep A) infections result in greater than 500,000 deaths annually. First line treatment for such infections is benzylpenicillin, often with the addition of clindamycin, but treatment failure can occur with this regimen. This failure has been partially attributed to the inoculum effect, which presents as reduced antibiotic susceptibility during high bacterial density and plateau-phase growth. Hollow fibre infection models (HFIM) have been proposed as an in vitro alternative to in vivo research to study these effects. To re-evaluate the inoculum effect for benzylpenicillin, clindamycin, linezolid, and trimethoprim-sulfamethoxazole using a Strep A HFIM. Differential antibiotic susceptibility of Strep A was measured in a HFIM starting from low- and high-density inocula with an average difference in bacterial concentration of 56-fold. Dynamic antibiotic concentrations were delivered over 48 h to simulate in vivo human pharmacokinetics in an in vitro model. Differences in antibiotic susceptibility were measured by plate count of colony-forming units over time. Inoculum effects were seen in benzylpenicillin and linezolid at 24 h, and benzylpenicillin, linezolid, and clindamycin at 48 h. The effect size was greatest for continuously infused benzylpenicillin at 48 h with a log-fold difference of 4.02 between groups. No inoculum effect was seen in trimethoprim-sulfamethoxazole, with a maximal log-fold difference of 0.40. Inoculum effects were seen using benzylpenicillin, linezolid, and clindamycin, which may predict reduced clinical efficacy following treatment delay. The model has proven robust and largely in agreeance with published data, recommending it for further Strep A study.