Comparison of In-Vitro and In-Vivo Drug Release of Dexamethasone/PLGA Intravitreal Implants Using Two In-Vitro Release Methods and a Rabbit Model

Drug delivery products based on poly(lactic-<i>co</i>-glycolic acid) (PLGA) are complex systems that may have significantly different drug release behavior under different release conditions, i.e., in different in vitro release tests (IVRTs) or in vivo. This must be appreciated when designing an IVRT for a given application. For example, when designing an IVRT for formulation development or bioequivalence testing, it is commonly desired that the IVRT is biorelevant to some extent, meaning that it imitates the relevant biological fluid/physiological environment ("compositional biorelevance") and by doing so simulates the in vivo release process ("mechanistic biorelevance"). Theoretically, biorelevance leads to "biopredictiveness," meaning that the IVRT predicts certain aspect(s) of in vivo performance. The objective of this study was to assess two IVRTs for dexamethasone/PLGA intravitreal implants according to a few selected aspects of biorelevance and biopredictiveness. These two IVRTs differed only in release medium composition: one used unbuffered isotonic saline, whereas the other used 12 mM phosphate-buffered saline (PBS) at pH 7.4. In the first part of this study, two different implant formulations were tested in these two IVRTs. The two implant formulations had the same composition and structure as each other but were manufactured using slightly different batches of PLGA. In the saline-based IVRT, the two formulations had similar in vitro release profiles, both lasting about 28 days; in the PBS-based IVRT, the two formulations had different release profiles, both lasting about 4 months. Next, these two implant formulations were injected into the left and right vitreous of 36 New Zealand White rabbits. In vivo drug release was measured directly by recovering implants at various timepoints and assaying their remaining drug content by high-pressure liquid chromatography (HPLC). Additionally, ocular pharmacokinetics were measured using liquid chromatography-tandem mass spectrometry (LC-MS/MS). The two implant formulations had similar in vivo release profiles and ocular pharmacokinetics as each other, and in vivo release lasted between 21 and 28 days; therefore, these in vivo results closely matched the saline-based IVRT. In the final part of this study, we endeavored to reconcile the saline-based IVRT's good prediction of in vivo release with the fact that the rabbit vitreous is buffered. This was investigated mechanistically by measuring the degradation of the implant's PLGA matrix during both in vitro and in vivo release. Matrix degradation was measured by recovering implant samples at various timepoints and measuring PLGA molecular weight by gel permeation chromatography. This was performed with one implant formulation in both IVRTs and in the vitreous of 4 rabbits. These measurements demonstrated that the PBS-based IVRT provided a better simulation of the implant's in vivo <i>surface</i> erosion, whereas the saline-based IVRT provided a better simulation of in vivo <i>bulk</i> erosion. Ultimately, we concluded that both IVRTs may provide value depending on the application: the saline-based IVRT may be preferred for predicting rabbit in vivo release, whereas the PBS-based IVRT may be preferred for a highly sensitive discriminatory test.