Sir

In an article on bioprospecting1, one of us (D. N.) was quoted to the effect that “only one sample in 250,000 will directly yield a commercial drug, although many more samples may serve as useful leads for modification through combinatorial chemistry”. It was not made clear that this estimate is based on data from the antibiotics industry2,3 and refers specifically to drug discovery from microbial sources. These data should not be translated directly to other natural sources, such as plants and marine organisms.

With estimates of the number of terrestrial plant species ranging from 250,000 to 400,000, the implication would be that only one or two commercial anti-cancer drugs might be developed from this source. This is an erroneous conclusion.

Indeed, the experience of the US National Cancer Institute (NCI) with the screening of plants between 1960 and 1982 indicates that some 35,000 samples (leaves, bark, roots and so on), derived from an estimated 12,000 species collected mainly from temperate regions, yielded two highly significant anti-cancer drugs, paclitaxel (Taxol) and camptothecin (which has been converted so far into two commercial products, topotecan and irinotecan), and homoharringtonine, which shows significant clinical efficacy against several leukaemias. This gives a success rate of one in about 4,000 species tested for activity in one disease category. A substantial improvement in the ‘hit’ rate can be expected with expanded screening.

We wish to emphasize that the NCI is continuing to explore the potential of nature as a source of novel pharmacophores in collaboration with the extramural scientific community. Where feasible, collection contracts are being replaced by direct collaboration with qualified organizations in the source countries, particularly in the investigation of plants.