Assessment of Differences in Antimicrobial Effect Determined with Two In Vitro Pharmacodynamic Models: Impact of Surface Area to Volume Ratio

Roger L. White, Pharm.D., Charles R. Bonapace, Pharm.D., Lawrence V. Friedrich, Pharm.D., Michael J. Rybak, Pharm.D., Diane M. Cappelletty, Pharm.D., Renee-Claude Mercier, Pharm.D., Heather H. Houlihan, Pharm.D., Jeffrey R. Aeschlimann, Pharm.D., John A. Bosso, Pharm.D.

Pharmacotherapy. 2003;23(5) 

In This Article


Numerous in vitro models, designed to mimic a variety of physiologic conditions, have been used to evaluate the antimicrobial activity of various drug regimens. Characteristics such as volume of central and peripheral compartments, number of compartments, and SA:V differed among them. Both one- and two-compartment in vitro models are frequently used to assess antimicrobial pharmacodynamics. Two-compartment in vitro models, in which the peripheral compartment is intended to represent the site of infection, differ in physical attributes of the central and peripheral compartments as well as in membranes that separate them. To date, no standards or guidelines are available for the physical nature (volumes, membrane composition, SA:V) or performance characteristics of in vitro models. An important difference among the models may be the SA:V of the peripheral compartment. Researchers using two-compartment in vitro models as well as rabbit models reported that the SA:V appeared to be an important determinant of the drug concentration-time profile in the peripheral compartment.[18,19,20] A smaller SA:V results in lower concentrations in the peripheral compartment that do not fluctuate to the same degree as those with large SA:Vs.[16,18,19] These differences would likely affect observed microbial killing patterns and thus calculated pharmacodynamic measurements, which often are used to characterize and compare antimicrobials in these in vitro systems.

We evaluated two similar in vitro models with different peripheral compartment SA:Vs. Although the magnitude of difference in SA:Vs between models was small (11%), differences in percentage of penetration and percent effect were large. Ceftazidime concentrations in the peripheral compartment of model A increased at a greater rate and reached maximum concentrations sooner than in model B. This resulted in a disproportionate increase of 53% in ceftazidime penetration into the peripheral compartment for model A compared with model B from 0-4 hours. However, over the 24 hours, peripheral compartment concentrations and percent penetration into that compartment were similar for both models. No significant differences were found in peak concentrations in peripheral compartments with SA:Vs of 3-10 cm-1; however, when the SA:V was 100 cm-1, the concentrations were significantly higher than those with smaller SA:Vs.[18] Although those results may appear to differ from ours, that study evaluated only drug concentrations after the last of several doses, whereas we evaluated both drug concentrations and their associated antimicrobial effects immediately after dosing, as is typically done in in vitro pharmacodynamic studies.

Ultimately, the increased penetration in model A resulted in a greater percent effect compared with model B. The effect was 64% greater from 0-4 hours and 38% greater from 0-24 hours in model A despite slightly higher percent penetration into the peripheral compartment in model B over 24 hours. These findings support the importance of early bacterial killing on the overall assessment of antimicrobial effects. Since measures such as the AUKC are an integrated measure of time and colony counts, significant killing early in the experiment without subsequent regrowth dominates calculation of the AUKC and thus the computed percent effect. It is possible that variations in growth rate may have contributed to observed differences between the models. Growth-control curves differed between the models, with model A reaching higher bacterial colony counts at 18 and 24 hours. Previous work suggested that with -lactams, which exert a greater bacterial effect against rapidly growing organisms, the rate of bacterial killing is dependent on the growth rate of the organism.[21,22]

Other confounding factors could have influenced our results, including the fact that the experiments were performed in different laboratories with slightly different experimental methods (microbiologic warm room vs heated water bath). However, a standard laboratory strain of P. aeruginosa (ATCC 27853) was used and the in vitro behavior of this organism is reproducible. Furthermore, the incubation temperature (37°C) was the same, and it is unlikely that differences in incubation would result in different microbial growth patterns.

Two additional issues deserve comment. First, this was a pilot project conducted with a single organism-antimicrobial combination. Differences in antimicrobial effects associated with differences in SA:V may be greater with other antimicrobials that have concentration-dependent bactericidal activity, such as aminoglycosides and fluoroquinolones. Observed effects also could differ among tested organisms. Second, differences in SA:Vs between models were relatively small. However, despite these relatively small differences, there were substantial differences in percentage of penetration and subsequently in percent effect, illustrating the importance of this property. Ideally, it would be preferential to study a much wider range of SA:Vs to elucidate its impact more fully.

Differences in experimental results among in vitro models may be due to many factors; however, differences in antimicrobial effects in these two models appear to be explained by differences in drug penetration into the peripheral compartment, which in turn are likely related to differences in effective SA:Vs. Future studies in in vitro models should fully describe the characteristics of the membrane and SA:V of the peripheral compartment. Furthermore, such characteristics should be considered in interpreting and comparing results. Characteristics of the peripheral compartment should mimic as closely as possible those that are thought to occur at the site of the infection in vivo.

Presented in part at the 38th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Diego, California, September 24-27, 1998.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.