The pharmacokinetic performance of a matrix system for transdermal beta-estradiol (E(2)) delivery after multiple consecutive dosing in postmenopausal women undergoing hormone replacement therapy was investigated. The E(2) plasma profiles determined during the third application in 16 postmenopausal women were compared with results obtained in a published clinical study using the same patch in 24 postmenopausal women without E(2) pretreatment; they were compared with a theoretical diffusion/pharmacokinetic model. A conventional theoretical model with constant model parameter (CPM) obtained from in vitro mass balance experiments in a Franz cell type set up described successfully the transdermal E(2) bioavailability parameter AUC(0-96h) (4341.9 +/- 1513.1; calculated 4250.8) and C(average) (45.0 +/- 13.2; calculated 41.2). Also, experimentally, there was no significant drop in E(2) plasma values after patch removal and reapplication; this was corroborated by calculations. Accumulation of E(2) did not occur when several patches were applied consecutively over a period of 3 weeks. Steady state was achieved following application of the first patch. However, the differences between recorded E(2) plasma profiles and theoretical results detected at specific measurement points cannot be explained by the CPM model. Experimentally obtained plasma profiles were always lower in the morning and higher in the evening than predicted on the basis of the model. Measurements of in vivo skin temperature in the postmenopausal women showed oscillating temperature profiles in the form of a cosinor function: The temperature mesor of untreated postmenopausal women was 34.8 degrees C with an acrophase at 17.0 o'clock (95% CI: 14.30-19.30) and an amplitude of +/- 0.4 degrees C (p = 0.1). During the application of the patch the average temperature next to a patch rose 0.3 degrees C, which was statistically significant (p = 0.1). In the skin under the application of the matrix patch a mesor temperature was detected as 35.6 degrees C with an amplitude of +/- 0.5 degrees C with an acrophase at 17.51 o'clock (95% CI: 14.30-21.00) (p = 0.05). The temperature period was 24 h for all measurements and the maximum temperature was observed at about 16.30 h, and a minimum at about 5.00 h. A linear dependency was detected in in vitro experiments between the log of E(2) permeability and the temperature for stripped skin, epidermis/dermis layer, as well as for the matrix. Modeling of E(2) plasma profiles with oscillating diffusion coefficients (ODM1) with a sine wave function results in this equation: D(1) = D(0x) + Da(x).sin(k.t). D(0x) is the diffusion coefficient determined at 35.6 degrees C, k is 1/24 h, D(a) is the diffusion coefficient of the temperature amplitude, h is hour, and x stands for the respective diffusion layer. It was shown that the experimental E(2) plasma profile variations are more pronounced than can simply be explained by skin temperature variations alone (ODM1 model). A simplex fit with an oscillating diffusion coefficient in the form of a sine wave function for the stratum corneum (ODM2 model) resulted in a temperature amplitude of 1.1 degrees C, about twice as high as was determined in the in vivo measurements (ODM2 model). Therefore, other circadian parameterlike blood flow might superimpose the temperature profile. The improvement in data analysis by incorporating oscillating diffusion coefficients (ODM1 and 2) over CPM was judged from a comparison of experimental data with the calculated plasma profiles with the AIC, Akaikes model selection criterion, which allows ranking between models because it is independent of the scaling of the data points. ODM1 and ODM2 improved the data analysis over CPM by allowing better calculation of experimental C(max), t(max), the time to reach to C(max), and the fluctuation, f. No difference between CPM, ODM1, or ODM2 was found for the bioavailability parameter C(average) and AUC(0-96h).
(c) 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association.