Comparative molecular field analysis and hologram quantitative structure activity relationship studies of pyrimidine series as potent phosphodiesterase 10A inhibitors

The inhibition of phosphodiesterase (PDE) 10A enzyme as an alternative approach is shown to have potential in the current treatment of schizophrenia. In the present work, the inhibition activity of 87 structurally diverse pyrimidine derivatives as PDE10A inhibitors were subjected to 2‐dimensional quantitative structure activity relationship (2D‐QSAR) and 3D‐QSAR studies using hologram QSAR and comparative molecular field analysis (CoMFA), respectively. The best CoMFA model gave a noncross‐validated correlation coefficient, (rncv2) = 0.977, SEtrain = 0.247 and predictive correlation coefficient (rpred2) = 0.957, and SEtest = 0.358. For the hologram QSAR (HQSAR) model, the similar parameters were rncv2=0.889, SEtrain = 0.565, rpred2=0.801, and SEtest = 0.872. The leave‐one‐out (LOO) cross‐validated r2 (q2) were 0.78 and 0.8 for the CoMFA and HQSAR models, respectively. These statistics indicate that the CoMFA model is more satisfactory than the HQSAR. Analysis of CoMFA steric and electrostatic contour maps strongly demonstrate that the presence of bulky group at the end of R1 substituent, especially heterocyclic aromatic rings, increase the inhibitory activities, and electronegative groups, such as alkoxide at the R2 position, enhance the inhibitory activities of studied chemicals. The results of this study serves as a guide for the possible modifications and designing of novel pyrimidine derivatives as new potent PDE10A inhibitors. The comparative molecular field analysis (CoMFA) model was successfully developed on 87 pyrimidine‐based derivatives as a PDE10A inhibitor. Analysis of steric and electrostatic contour maps of the derived CoMFA model suggests sufficient information to improve the inhibitory activity by modification of the structure of the studied inhibitor compounds. The constructed CoMFA model demonstrates statistically robust results with a q2 of 0.782, r2ncv of 0.977, and a high predictive ability for the test set (r2pred = 0.957).