Information Technology Reference
In-Depth Information
polymers, Eudragit RS-PM and Ethocel 100, with the fi nal goal of
drug release optimization (Sánchez-Lafuente et al., 2002). Four
independent variables were considered: compression force used
for tableting, ratio between the polymers, their particle size, and
drug content. The considered responses were the percentage of
drug released at 3 predetermined times, the dissolution effi ciency
at 6 h, and the time to dissolve 10% of the drug. These responses
were selected, since the percentage of drug released after certain
time points is considered the key parameter for any
in vitro/in vivo
correlation process. The preliminary screening step, carried out by
means of a 12-run asymmetric screening matrix according to a
D-optimal design strategy, allowed evaluation of the effects of
different levels of each variable. Different levels of each
independent variable on the considered responses were studied:
compression force and granulometric fractions of polymers were
varied on 2 levels, drug content was varied on 3 levels, and the
ratio between the polymers was varied on 5 levels. Starting from
an asymmetric screening design 2
2
3
1
5
1
//18, D-optimal design
strategy was applied and a 12-run asymmetric design was
generated. The drug content and the polymers ratio had the most
important effect on drug release, which, moreover, was favored by
greater polymers particle size; to the contrary, the compression
force did not have a signifi cant effect. The Doehlert design was
then applied for a response-surface study, in order to study in-depth
the effects of the most important variables. In general, the
Doehlert design requires
k
2
+
k
+
n
experiments, where
k
is the
number of factors and
n
the number of central points. Replicates
at the central level of the variables are performed in order to
validate the model by means of an estimate of the experimental
variance. In this study, drug content was varied on 3 levels and the
polymer ratio was varied on 5 levels. Response surfaces were
generated and factors interactions were investigated. The
desirability function was used to simultaneously optimize the 5
considered responses, each having a different target. This
procedure allowed selection, in the studied experimental domain,
Search WWH ::
Custom Search