Crystal structure prediction method successfully models axitinib polymorphs and salt formation
A groundbreaking study has demonstrated the power of crystal structure prediction in unraveling the complexities of axitinib, a targeted anti-cancer medication. Gregory Beran, a researcher at the University of California Riverside, has achieved a significant milestone by combining hybrid density functional theory modeling with intramolecular energy correction, enabling the reliable prediction of axitinib's intricate crystal structures. This breakthrough not only predicts the most stable polymorphs but also distinguishes between salt and co-crystals in multi-component crystals, a long-standing challenge in organic crystal modeling.
Axitinib, prescribed to patients with advanced renal cell carcinoma, functions as a tyrosine kinase inhibitor, blocking vascular endothelial growth factor receptors that drive tumor growth and metastasis. However, its development has been complicated by its ability to crystallize in multiple forms, known as polymorphism. Axitinib can exist in five known non-solvated crystalline structures, each with distinct physical and chemical properties that influence solubility and bioavailability. Initially, chemists thought form IV was the most suitable candidate, but further experimental screenings revealed form XXV and later form XLI, which is the most thermodynamically stable and ultimately received US Food and Drug Administration (FDA) approval in 2012.
Beran's research highlights the potential of crystal structure prediction to avoid such pitfalls. By combining density functional theory (DFT) with intramolecular energy correction, he has produced the first 0K crystal structure prediction of axitinib that aligns closely with experimental data. He optimized several known crystal structures of axitinib, predicting a version of form IV with two independent molecules in its unit cell and multi-component crystals with fumaric, suberic, and trans-cinnamic acid.
The study's methodology involved applying periodic planewave density functional theory and exchange-hole dipole moment dispersion correction to relax crystal structures. Constrained optimizations mapped energy curves for proton transfer coordinates in multi-component crystals, and crystal energies were refined using periodic density functional theory with generalized-gradient approximation and hybrid functionals. Intramolecular correction refined single-point periodic density functional theory energies, and Orca software performed polarisable continuum model treatment for low- and high-level calculations.
The crystal structure prediction revealed that most low-energy predicted structures of axitinib are known experimentally, and form XLI is the global minimum structure at 0K. The study also distinguished the axitinib salt formed with fumaric acid from co-crystals involving suberic or trans-cinnamic acids.
Sarah (Sally) Price, a theoretical chemist at University College London, emphasizes the study's potential, stating that it points towards a methodology for predicting whether two molecules would crystallize as a salt or co-crystal. This opens up exciting possibilities for modeling experimental data in the computer. Beran acknowledges the progress made but also acknowledges that crystal structure prediction is still not perfect and requires further improvements as a community.
