Elucidation of the molecular defects in cancer and anticancer drug development should, in theory, go hand in hand. Significant drawbacks can occur when drug development occurs before the biology has been worked out.

This topsy-turvy approach is illustrated in the review on page 657 by Christopher Overall and Carlos López-Otín. They discuss recent insights into the mechanisms by which matrix metalloproteinases (MMPs) contribute to tumour growth. This knowledge has not only helped to explain why first-generation MMP inhibitors have been disappointing in clinical trials, but will also help to develop new therapeutic strategies to target MMPs. New work has also helped our understanding of the role of MMPs and indicates that a stromal-therapy approach might be the next logical step (see Highlight on page 637). The need for increased communication between basic scientists, translational researchers and clinicians has never been more relevant.

In line with this need, we introduce a new feature in this issue — From the analyst's couch — that we will reprint from Nature Reviews Drug Discovery when they cover a cancer-related subject. The feature on page 645 summarizes the properties of several selective anticancer drugs and looks ahead to the impact that these drugs are likely to have on patients and the drug market. In the examples given, unravelling the details of the relations between chosen targets and the disease were done in parallel with drug development.

Although rational drug design might be the ideal, often a chance finding can lead to success. In the case of arsenic trioxide (see the Timeline by Jun Zhu et al. on page 705), modern biological techniques have led to the rediscovery of an ancient Chinese medicine, arsenic, as a potent treatment for acute promyelocytic leukaemia.