| Cephalosporins have made a progress in this century since discovered, and there are more than 50 kind half-synthesized cephalosporins on world market. The forth cephalosporin antibiotics have been developed which have ranked first in antibiotic field. In recent years, for the unceasing rising incidence of Haemophilus influenzae disease, people pay more attention to it. Except for a few of cephalosporins having high antibacterial activity, almost all had low activity against Haemophilus influenzae. Moreover cephalosporin could redound to many toxicity and side effect along with its widely clinical application such as immune response, congnitive malfunction of haematopoietic system, subsequent infection, liver and kidney toxicity and so on. It was urgent for us to discover new cephalosporin derivants having not only excellent antibacterial activity but also good pharmacokinetic quality. Therefore the research objective was to analysis the relationship between cephalosporin structure and pharmacodynamics/pharmacokine-tics, then to put forward some improvement opinion about changing cephalosporin structure, and offer some references and help for designing new cephalosporin compounds. In this paper, the research on QSAR/QSPR for cephalosporins by use of quantum chemistry and neural network methods was made. 102 cephalosporins were selected and their minimum inhibition concentrations to Haemophilus influenzae had been determined. The 24 molecule structure parameters had been calculated by quantum chemistry AM1 algorithm. The correlative analysis of the 24 structure parameters was analyzed and 7 structure parameters were selected as input parameters of neural network. A QSAR model for cephalosporins by BP neural network was designed and Levenberg-Marquardt algorithm was used for training. By this network, 94 compounds were selected stochastically as the training set and the residual 8 ones as the test set. The results showed the relative errors of being learned and tested of the minimum inhibition concentrations to Haemophilus influenzae were small. It showed that the selected structure parameters and QSAR model for cephalosporins were relatively reasonable, on certain level, could embody antibacterial activity against Haemophilus influenzae of these compounds. On the other hand, 26 compounds were selected to study the relationship between the molecule structure parameters of cephalosporins and PK parameters( t1/2, t1/2β, AUC, CL and Vap). Similarly, the 18 molecule structure parameters of the 26 compounds were calculated and 6 parameters were choosed for input of the net. 23 compounds were selected stochastically as the training set and the residual 3 ones as the test set. The results showed the relative errors of learned and tested values were small in QSPR model. All the results showed the QSPR model between the pharmacodynam-ic parameters and the structure parameters of cephalosporins was reasonable. At present the main work to design new cephalosporin compounds was how to change C-7 side chain and C-3 side chain and keep 7-Amino-cephalosporanic acid structure in order to improve curative and antibacterial activity, extend antimicrobial spectrum, improve stability against β-lactamases, reduce toxicity and change their pharmacokinetic quality. On base of above analysis, 30 new cephalosporin compounds were designed. Utilizing the above QSAR and QSPR model to predict their antibacterial activity against Haemophilus influenzae and their pharmacokinetic parameters, there were two compounds had both high antibacterial activity against Haemophilus influenzae and good pharmacokinetic quality, twocompounds had some antibacterial activity to some degree but their pharmacokinetic quality was very good. All these work offered a new way of thinking and references to design the new cephalosporin drugs. |
