(351a) System Identification and Frequency Response Techniques for the Design of Controlled Release Drug Delivery Systems

System
Identification and Frequency Response Techniques for the Design of
Controlled Release Drug Delivery Systems

Timothy Knab, Sam
Rothstein, Steven R. Little, and Robert S. Parker

Department of Chemical and
Petroleum Engineering

University of Pittsburgh,
Pittsburgh, PA 15237

There
is great interest in developing truly programmable controlled release
systems and, to that end, numerous models have been developed.
However most models are not broadly predictive and fewer still allow
for direct correlation of release behavior to the controllable
physical properties of the device. To address this issue, our group
has recently produced a model for well-defined, programmable
controlled release in vitro^[1],
which to our knowledge, can be applied to more drugs and polymers
than systems described to date. However, the current model is limited
to an idealized in vitro environment and further work is required to
account for the non-ideal in vivo processes that may impact release
behavior. Our goal is, for the first time, to extend the current
release model to have predictive capabilities in an in vivo
environment and systematically determine the most important
parameters governing release in vivo. We hypothesize that tissue
environment-specific effects, affecting release and transport rates,
can be detected in systemic circulation and characterized using
system identification techniques, namely frequency response analysis.
Using the predictive power of the aforementioned model, we can
synthesize particles that yield specific release profiles designed to
inform the data-driven analysis.

As
a basis for this study, a proof of concept diffusion experiment
has been developed. Changes in membrane diffusivities in a
simulation of this experiment lead to changes in reservoir
concentrations over time, and these differences produce identifiable
changes in the frequency response characteristics as evidenced from
Bode plots. This type of analysis requires a well-controlled input
function -- taken from the mathematical model of controlled release
- and an output signals that is easily measurable. For this we use
flux, which is easily attainable from an actual experimental
measurement of concentration via a derivative relationship. Over a
two order of magnitude span in diffusivities these simulations show
these plots changing from flat, indicating no lag, (rapid diffusion)
to a rapid drop off in the magnitude and phase indicating slow
diffusion and equilibration of the second tank.

Experimental
studies on the physical realization of the simulated diffusion cell
are being compared against the simulated data as a test/validation of
the experimental system and the numerical tools. Although the
dynamics of diffusion cells are well studied, this system can be
extended to be more analogous to in vivo physiology through
experimental modifications, such as using an ECM seeded with cells as
a barrier to transport or the introduction of a diffusive barrier
mimicking the effects of fibroblast encapsulation of foreign
microparticles. If successful, this system would provide a design
framework for controlled release where exquisite control of the
release system can be used to characterize -- and overcome --
potential barriers to drug release. The result is the ability to
"pre-program" a microparticle for a desired release profile
that results in a specified concentration profile at a location
remote to the particle.

The
system identification techniques we propose to use require reliable,
high resolution tracking of spatially disparate drug or tracker
molecule concentrations. To that end,
Gadolinium-tetraazacyclododecanetetraacetic acid (Gd-DOTA) is being
investigated as a potential tracking agent. Gd-DOTA concentrations at
various locations within a system can be directly related to magnetic
resonance imaging (MRI) T1 relaxation times. The availability of a
non-invasive non-destructive imaging platform facilitates the
translation of our tools to spatio-temporal analysis of experimental
and in vivo systems. Initial work has focused on the design and
release of Gd-DOTA from microparticles and the characterization of
the diffusion profile in a polymer gel matrix. Gd-DOTA seems to be
capable of coordinating with the microparticle polymer matrix, which,
in effect, acts as a ligand and results in extremely delayed release
compared to the model-simulated release that neglects this
drug-polymer coordination. This is thought to be due to the interplay
between intra-microparticle pH and Gd-DOTA charge -- an observation
consistent with protein release studies also taking place in our lab.
For our techniques to be successful, the effects of pH and charge
needed to be included in the current model. We are currently working
on a mechanistic description of these effects that will ultimately
lead to an even more broadly predictive model for controlled release
that can also be used in the development of a model of
pre-programmable, controlled release, in vivo.

[1] Rothstein,
S. N., Federspiel, W. J., & Little, S. R. (2008). A simple model
framework for the prediction of controlled release from bulk eroding
polymer matrices. Journal of Materials Chemistry, 18(16),
1873. doi:10.1039/b718277e