Near-molecular Resolution for Drug Formulation Studies

David England from Sanofi-Synthelabo has used DPD simulations to rationalize the behavior of complex polymer-based drug formulations and help to understand changes which occur during stability studies.

Such changes impact the suitability for administration to patients and must be determined.

Enables the early elimination of candidates that can’t be administered to patients.

Improved quality and made drug development more efficient.

The full characterization of a drug product is important in the understanding of how it will perform during administration and how it will change during its shelf life. The results of such characterization studies, however, can be difficult to rationalize. This is particularly true for polymer-based formulations where mesoscopic scale (10-1000 nm) changes in crystallinity, e.g. the formation of lamellae phases for polyethylene oxide, and phase separation can occur, e.g. semi-solid formulations. Such changes may affect the appearance of the product or cause crystallization of the drug from the matrix. Molecular modeling combined with physicochemical analytical studies have an important role to play in understanding these changes.

In such a formulation, comprising of a mixture polyethylene glycol (PEG400), a triblock copolymer of ethylene oxide-propylene oxide-ethylene oxide (Poloxamer407), and Polysorbate20, atomistic simulations have been used to calculate energies of mixing for the various chemical species in the system. Coarse-grained models of each of these molecules have then been developed and DPD (dissipative particle dynamics) has been used to follow the time evolution of interactions using a short-ranged soft potential, which is proportional to the energy of mixing for the pair of beads in question.

Here the color coding indicates how the drug molecule can be simplified for coarse-grained simulations.
At a coarse-grained level, the molecule is simply a group of beads. The colors show that the drug comprises beads of various types, which will have different interactions with the other components of the formulation. The interaction of the beads with each other can be calculated using atomistic simulations of the interactions of the substructures to generate a simplified forcefield.

This approach should then be used to coarse-grain all components within the formulation.

Poloxamer407. The ethylene oxide chains (gray) are attached to a central block of propylene oxide (green).
PEG400. This can be modeled as 3 beads each equivalent to 3 monomer units.
Polysorbate20. A branched molecule with predominantly ethylene oxide components (gray beads), but one branch is capped with an ester group (blue bead).

The components can then be mixed in the ratio used for the formulation to develop a simulation, which investigates the interaction of many thousands of molecules.

A DPD simulation was started from a homogeneous mixture with the following composition:

Molecule	Concentration (%)
Drug	8
PEG400	46
Poloxamer407	18
Poloxamer407	18

Starting from a randomly mixed model, after equilibration the system self-assembles. Here is a snapshot from the endpoint of the simulation. The pink surface links points where the drug density falls to half of its maximum. The blue surface is the same for the polypropylene oxide block of the poloxamer.

It should be noted that this is a single snapshot from a dynamics system. These profiles show that a shell of drug forms around the propylene oxide rich regions; this is driven by the hydrophobic nature of the drug. We can rationalize the above simulation image with the following schematic:

Above: Micelle type objects are formed with a corona comprising of ethylene oxide chains. The core is propylene oxide surrounded by a layer of drug. The matrix is PEG400 and Polysorbate20.

The simulation can be taken to the next stage, to simulate dissolution. When water is allowed to interact with the preformed formulation, the Polysorbate20 (green), which has surfactant properties, coats the core.

A simplified description of the dissolution system.

During stability studies it was observed that the above formulation when manufactured with vigorous stirring during cooling produced a clear product. If however only gentle stirring was used a white product formed which over approximately 6 months became clear in appearance. Imaging of the formulation during cooling indicates that the source of the opacity is crystallization of the polymer components of the formulation.

Spherulites formation is a typical consequence of the crystallization of polyethers, PEG, and Poloxamer. DPD was also used to rationalize this behavior.

The effect of processing was mimicked in the DPD simulations by starting a dynamics run from two distinct initial states: (1) Completely mixed (to represent the vigorously agitated system as discussed above) and (2) with a degree of phase separation of the individual components. The early stages of the second simulation are shown below: even after long simulation we see only incomplete wetting of the drug by the polymer.

The orange regions are drug rich and the blue are the poloxamer polypropylene oxide regions. The separation of these regions is long-lived on the DPD timescale, suggesting that this would facilitate crystallization of the polymeric components.

Since this demixed state will facilitate crystallization of components it is not surprising that the appearance of this formulation changes over time as this metastable product eventually develops towards the thermodynamically stable form.

Conclusion

DPD simulations have been used to rationalize the behavior of complex polymer-based drug formulations and help to understand changes that occur during stability studies, such that their impact on the suitability for administration to patients can be assessed.