Abstract
Background
The relationship between antimicrobial activity and exposure relative to MIC is typically evaluated using one of three PK-PD indices, AUC:MIC ratio, Cmax:MIC ratio, and %T>MIC. However, under certain circumstances, none of these PK-PD indices may be the most optimal. These include when the fitted Hill functions for each of the PK-PD indices do not allow for sufficient discrimination, the variability about the fitted functions is wide, and/or the pattern of dose fractionation data is non-informative. Relationships fit using the traditional PK-PD indices may be suboptimal for drugs which exhibit extreme PK characteristics such as abnormally short or long half-lives. As described herein, we explored the use of a fourth PK-PD index for such instances, AUC/τ:MIC ratio (τ = dosing interval).
Methods
Previously-described ceftolozane dose-fractionation data from a study using a neutropenic murine thigh-infection model were evaluated [AAC 2013; 57(4):1577–82]. In this prior study, mice were infected with E. coli ATCC 25922 (MIC = 0.5 mg/L) or K. pneumoniae ATCC 43816 (MIC = 1.4 mg/L). Ceftolozane doses ranged from 1.56 to 1600 mg/kg/24h given q3h, q6h, q12h, or q24h. Relationships between log10 colony forming units (CFU) at 24 hours and AUC:MIC ratio, Cmax:MIC ratio, %T>MIC, and AUC/τ:MIC ratio were evaluated by pathogen and pooled using Hill-type models and non-linear least squares regression.
Results
For evaluations of data by pathogen, AUC/τ:MIC ratio best described changes in log10 CFU at 24 hours. The coefficients of determination (r2) for these pathogens were improved by 0.20 and 0.11, respectively, relative to the highest r2 achieved using any of the traditional PK-PD indices. Similar results were observed when the data were evaluated using a pooled approach (Figure 1).
Conclusion
AUC/τ:MIC ratio may be useful to evaluate drugs demonstrating the extremes of PK. Accordingly, this PK-PD index best described ceftolozane PK-PD, an agent with a very short murine plasma half-life (<15 minutes). The use of the PK-PD index that allows for the best fit of the data to the Hill function and reduced variability about the fitted function will not only improve the characterization of PK-PD but will also improve the accuracy of future dose selection analyses.
Disclosures
All authors: No reported disclosures.
Quality characterization of wine cork stoppers using computer vision