A novel PD model for gatifloxacin (GF) vs. salmonella typhi (ST) in timed kill curves (KC)

Quinolones are the mainstay of treatment for ST infections but there is growing concern about resistance to older quinolones & poor understanding of quinolone/ST PD. We have developed a new approach to modelling KC data & have applied it to describing the PD of GF vs. ST. Log‐phase cultures (10 7–8 CFU/mL) of ST (MIC=0.5 mg/L) were exposed to GF at 0, 0.5, 1, 2, 4 & 8xMIC; bacterial counts (CFU) were measured serially over 24h. Timecourse of CFU were fit to a PD model with capacity‐limited bacterial growth, 1st‐order rate constant for death (Kd) & a Hill‐type function in which GF enhances Kd. The total CFU was represented by a mixture model, with up to 4 sub‐populations (SP) differing in GF susceptibility. Each KC was fit individually & later simultaneously, using Adapt II. AIC was used to determine the number of SP & which parameters would be allowed to vary between KC. The final model had inter‐KC variance in maximum velocity of growth & 4 SP. The 1st 2 SP were 99.99% of the initial CFU with sensitivities <1xMIC & the other 2 SP were each <0.01% of total CFU, with sensitivities of 2.3 & 4.4xMIC. The GF Emax was a 37‐fold increase in Kd. Goodness of fit was excellent with an overall r=0.96 (Observed=1.01Fit−0.15). This approach to PD modelling of in vitro data will give better insight into the activity of GF vs. different ST SP & aid in determining regimens which minimize therapeutic failure due to the development of resistance. Clinical Pharmacology & Therapeutics (2004) 75 , P42–P42; doi: 10.1016/j.clpt.2003.11.159