Polymorph landscape

Cheung VG. Polymorphic landscape of the human genome. EurJHum Genet 2005 13 133-135. 

Computational assessment of the likelihoods of occurrence and the relative stabilities of polymorphs is not necessarily more effective than the experimental approach. Whilst great advances have been made in the field of ab initio crystal structure prediction (CSP), as documented in five international blind tests spanning the years 1999-2010 [5], it is still not routinely possible to predict whether a molecule is likely to be polymorphic or to confirm whether the most thermodynamically stable structure has been found experimentally, especially for molecules of the complexity of a typical drug. It is possible to compute the polymorph landscape for a specific flexible molecule, but the calculations require considerable expertise, and the timescales and computing resources can render CSP impractical for application to even a limited portfolio of candidate APIs. 

As described earlier, the development time for generating a tailored force field used in CSPs is rather lengthy, and thus such an approach is not possible to use for each compound that a pharmaceutical company is interested in knowing the solid-state properties of [11-13]. Therefore, at AstraZeneca we have over the recent years developed a less time-consuming approach, which is here briefly described [14]. Rather than creating a new force field for a specific molecule, we have relied on the OPLS2005 force field and based on that created an in-house-modified force field— here called Solid State Prediction Tool (SSPT) for the treatment of the intermolecular interactions and for the creation of polymorph landscapes [15,16]. Our calculations typically follow the procedures listed down ... 

A typical calculated polymorph landscape may look as in Figure 7.2 as generated for benzamide, with calculated interaction energies plotted versus the Kitaigorodsky packing index [20]. As a reminder to the reader, the most stable polymorph is the calculated crystal structure with the most negative interaction energy. However,... 

FIGURE 7.2 Predicted polymorph landscape for benzamide. The two conformers investigated are illustrated by diamonds and circles. 

One may now ask how polymorph landscape investigations are useful for risk assessments for drug projects. A valuable piece of information is the density of predicted polymorphs, as illustrated in Figure 7.10. In the top plot is shown a polymorph landscape with few solutions and the most stable polymorph clearly separated from the other. This is an ideal situation where the predictions indicate a low risk for other polymorphs to be formed. In the lower plot, on the other hand, we see many predicted polymorphs with similar energy, thus a high density in the most interesting area of... 

FIGURE 7.10 Predicted polymorph landscapes, illustrating different scenarios with low-(upper) and high risk (lower), respectively, for formation of several stable polymorphs. 

FIGURE 7.12 Polymorph landscape of GPR119-2 generated with the SSPT CSP methodology. Plotted is the interaction energy (El in kJ/mol) versus packing index (ck in %). The crystal structure in the left-hand upper comer is the energetically most favorable structure. 

Now, if the rankl -rot conformation is used in a rigid CSP, the polymorph landscape is still busy, but a few solutions are clearly separated from the main group, see... 

At this point, we can conclude that the fully unbiased method fails as only small differences between the theoretical conformations and the observed will translate into a too complex polymorph landscape to allow for systematic reranking using DFT-D calculations on the tentative crystal structures. However, in this case we know the answer, and from all the predicted crystal structures we can pick... 

Levy Y, Becker OM (2002) Conformational polymorphism of wild-type and mutant prion proteins energy landscape analysis. Proteins Struct Funct Genet 47 458... 

Price, S.L. Computed crystal energy landscapes for understanding and predicting organic crystal structures and polymorphism. Acc. Chem. Res. 42, 117-126 (2008)... 

There are several well-documented cases of the conversion of existing marketed drugs to previously unknown polymorphs, for example ritonavir [2] and rotigotine [3], with serious medical, social and financial consequences. It is therefore crucial for drug development scientists to understand, as far as possible, the solid form landscape. 

The CSD-related scientific and software tools developed for polymorph risk mitigation, and cocrystal design, are the central focus of this chapter. We begin with a brief summary of the CSDS, and then discuss (i) the development and application of H-bond propensity analysis, (ii) the study of H-bond landscapes and (iii) informatics-based cocrystal screening. In each case we provide case studies to exemplify the methodology. Ongoing development areas and new opportunities are noted in the section Conclusions and Outlook . 

Thus far we have demonstrated the relationship of H-bond propensity, optimal inter-molecular bonding and the likelihood of polymorphism. Combining propensities with models of how many H-bonds may be formed by a given atom in a functional group (i.e. the H-bond coordination environment) allows the in silico generation of all chemically reasonable structures. Comparing a given solid form to other possible structures in the resultant H-bond landscape provides a powerful analysis of whether... 

---