To the editor:

Your editorial "Pharmacogenomics at work" (Nature Biotechnology 16:885, October 1998) is off base. The use of a diagnostic test that detects overexpression of the HER2 antigen to identify breast cancer patients likely to benefit from Herceptin is not an example of pharmacogenomics.

Proponents of pharmacogenomics argue that the genotype can be used to segregate responding from non-responding patients.

A test for HER2 overexpression measures a phenotype that is not linked to genotype. Conceptually the HER2 test resembles tests for estrogen receptors. Different diagnostics unrelated to genotype are common and should not be confused with speculation regarding potential value of identifying patient populations by genotype.