Interpreting the Physicochemical Meaning of a Molecular Descriptor Which Is Predictive of Amorphous Solid Dispersion Formation in Polyvinylpyrrolidone Vinyl Acetate

Abstract

A molecular descriptor known as R3m (the R-GETAWAY third-order autocorrelation index weighted by the atomic mass) was previously identified as capable of grouping members of an 18-compound library of organic molecules that successfully formed amorphous solid dispersions (ASDs) when co-solidified with the co-polymer polyvinylpyrrolidone vinyl acetate (PVPva) at two concentrations using two preparation methods. To clarify the physical meaning of this descriptor, the R3m calculation is examined in the context of the physicochemical mechanisms of dispersion formation. The R3m equation explicitly captures information about molecular topology, atomic leverage, and molecular geometry, features which might be expected to affect the formation of stabilizing non-covalent interactions with a carrier polymer, as well as the molecular mobility of the active pharmaceutical ingredient (API) molecule. Molecules with larger R3m values tend to have more atoms, especially the heavier ones that form stronger non-covalent interactions, generally, more irregular shapes, and more complicated topology. Accordingly, these molecules are more likely to remain dispersed within PVPva. Furthermore, multiple linear regression modeling of R3m and more interpretable descriptors supported these conclusions. Finally, the utility of the R3m descriptor for predicting the formation of ASDs in PVPva was tested by analyzing the commercially available products that contain amorphous APIs dispersed in the same polymer. All of these analyses support the conclusion that the information about the API geometry, size, shape, and topological connectivity captured by R3m relates to the ability of a molecule to interact with and remain dispersed within an amorphous PVPva matrix.

Machine Learning Study of Metabolic NetworksvsChEMBL Data of Antibacterial Compounds

“…Molecular descriptors enable the indirect correlation of desired attributes with a molecule’s structure . Analysis of the structural interpretation of the IFPTML-LDA model showed that total and local linear indices (atom and atom type) are the most influential descriptors in the chemical datasets, specifically nonstochastic matrix orders 2 and 3 in the presence of C atoms in the aliphatic chain.…”

HATS5m as an Example of GETAWAY Molecular Descriptor in Assessing the Similarity/Diversity of the Structural Features of 4-Thiazolidinone

“…The greater the atomic mass of a halogen, and hence its weight, the more significant its contribution to the descriptor value. Similarly, in the case of the R3m descriptor (R-GETAWAY third-order autocorrelation index weighted by the atomic mass), the presence of atoms heavier than C in a molecule results in a greater contribution to the R3m value [ 8 ]. Note, according to Equation (7), each atom’s contribution to HATS5m depends on both the atom weight (atomic mass in this case) and the leverage.…”

“…Consonni and co-workers [ 6 ] proposed a new molecular representation calculated from the spatial coordinates of the molecule atoms in a chosen conformation, allowing for a more accurate similarity/diversity analysis of compounds stored in databases. The development of structure–activity modeling of different biological/physicochemical endpoints confirmed the usefulness of GETAWAY descriptors in molecular sciences [ 8 , 9 , 10 , 11 , 12 , 13 , 14 ]. Several studies prove that the combination of GETAWAY and WHIM descriptors significantly improves the predictive power of the QSAR model [ 15 , 16 , 17 , 18 ].…”

“…This class of molecular descriptors can be sub-classified based on the matrix type: H-GETAWAY (molecular influence matrix H) and R-GETAWAY (influence/distance matrix R). A good example of a thorough interpretation of the R-GETAWAY third-order autocorrelation index weighted by the atomic mass descriptor (R3m) was recently published [ 8 ]. Unfortunately, such a detailed interpretation of descriptors in QSAR models appears relatively rarely in the studies.…”

Implications of Coexistent Halogen and Hydrogen Bonds in Amorphous Solid Dispersions on Drug Solubility, Miscibility, and Mobility

“…8,11,13,14 Typically, hydrogen bonding between drug and polymer makes ASDs more physically stable; 6,8 however, in the case of polyvinylpyrrolidone vinyl acetate random copolymer (PVPVA), molecules such as itraconazole and clotrimazole still form persistent molecular dispersions despite being unable to form hydrogen bonds with the carrier polymer. 15,16 Polymer screening is the first step in preclinical ASD development, 9,17,18 where ruling out drug−polymer compatibility based solely on hydrogen bonding potential overlooks other specific interactions, such as halogen bonds, which may lead to viable formulations. 19 More than 25% of approved FDA drugs contain at least one halogen atom.…”