Aim: To build up a quantitative structure-activity relationship (QSAR) model of
20 (S)-camptothecin (CPT) analogs for the prediction of the activity of new CPT
analogs for drug design. Methods: A training set of 43 structurally diverse CPT
analogs which were inhibitors of topoisomerase I were used to construct a quan-
titative structure–activity relationship model with a comparative molecular field
analysis (CoMFA). The QSAR model was optimized using partial least squares
(PLS) analysis. A test set of 10 compounds was evaluated using the model.
Results: The CoMFA model was constructed successfully, and a good cross-
validated correlation was obtained in which q2 was 0.495. Then, the analysis of
the non-cross-validated PLS model in which r2 was 0.935 was built and permitted
demonstrations of high predictability for the activities of the 10 CPT analogs in
the test set selected in random. Conclusion: The CoMFA model indicated that
bulky negative-charged group at position 9, 10 and 11 of CPT would increase
activity, but excessively increasing bulky group at position 10 is adverse to inhibi-
tory activity; substituents that occupy position 7 with the bulky positive group
will enhance the inhibitive activity. The model can be used to design new CPT
analogs and understand the mechanism of action.