Background
C-reactive protein (CRP) pharmacodynamic (PD) models have the potential to provide adjunctive methods for predicting the individual exposure-response to antimicrobial therapy. We investigated CRP PD linked to a vancomycin PK model using routinely collected data from non-critical care adults in secondary care.
Methods
Patients receiving intermittent intravenous vancomycin therapy in secondary care were identified. A two-compartment vancomycin PK model was linked to a previously described PD model describing CRP response. PK and PD parameters were estimated using a Non-Parametric Adaptive Grid technique. Exposure-response relationships were explored with vancomycin area-under-the-curve (AUC) and the index, AUC:EC50, fitted to CRP data using a sigmoidal Emax model.
Results
Twenty-nine individuals were included. Median age was 62 (21-97) years. Fifteen (52%) patients were microbiology confirmed. PK and PD models were adequately fitted (r2 0.83 and 0.82 respectively). There was a wide variation observed in individual Bayesian posterior EC50 estimates (6.95-48.55mg/L), with mean (SD) AUC:EC50 of 31.46 (29.22). AUC:EC50 was fitted to terminal CRP with AUC:EC50 >19 associated with lower CRP value at 96-120 hours of therapy (100mg/L vs. 44mg/L; p<0.01).
Conclusion
The use of AUC:EC50 has the potential to provide in-vivo organism and host response data as an adjunct for in-vitro MIC data, which is currently used as the gold standard PD index for vancomycin therapy. This index can be estimated using routinely collected clinical data. Future work must investigate the role of AUC:EC50 in a prospective cohort and explore linkage with direct patient outcomes.