Principles of antibiotic tissue penetration and guidelines for pharmacokinetic analysis

No one type of model can be clearly identified as the standard for tissue penetration studies. Each has their utility, and a prudent clinician will employ results from as many as are available. Even in the absence of definitive data, it may still be possible to propose some reasonable guidelines for determination of relevant tissue penetration data. The authors would generally recommend evaluating tissue penetration at the steady state, by direct measurement of the tissue concentration. Dividing this steady state tissue concentration by the steady state serum concentration yields a realistic partition ratio. Although this steady state tissue concentration determination also measures some drug in the blood, it is reasonable to use such numbers because an infection also involves both blood and tissue sites. However, one must also be cautious about applying values derived from normal tissues to the disease itself. There are no human or animal studies that clearly establish the clinical relevance of tissue concentrations. Such relevance depends on correlating tissue concentrations with cure of infection at the tissue sites. It is possible that the concentration in blood is as useful as tissue concentration in the prediction of cure, yet even studies of serum concentration versus cure are rare. Thus, the prudent clinician should remain quite skeptical of data that compare tissue penetrations of similar antibiotics and suggest that there are differences in efficacy. An antibiotic with better tissue penetration may or may not be more effective. For the clinician who would use tissue concentrations, the types of studies of highest potential value are those that measure the antibiotic concentration at all sites, determine the in vitro characteristics of the organism, and then find the correlations predictive of cure. Until such studies are performed, useful data on steady state partitioning, binding, and the effects of disease on these factors should be sought and carefully stored for later application in predictive models and systems.