T he U.S. Food and Drug Administration ’ s (FDA) approval of vemurafenib (Plexxikon/Roche) last August was greeted as a milestone not only for melanoma treatment but also for personalized medicine. The drug is indicated for patients who harbor a BRAF V600 → E mutation. Vemurafenib was developed along with a companion diagnostic, in a test case for how the FDA plans to handle personalized drugs from now on. In a draft guidance issued last July 14, the FDA defined companion diagnostics as devices that provide essential information for the safe and effective use of a corresponding therapeutic product. It stated that it would review personalized drugs only in tandem with the diagnostic that was used to select patients for clinical trials, adding that personalized drug labels should refer specifically to approved diagnostics for use in selecting patients for treatment. FDA offi cials emphasize that the approach is needed to ensure that personalized drugs are given only to likely responders. But as the experience with vemurafenib shows, the new method encroaches on entrenched diagnostic turf. Hospitals, drug companies, and professional medical societies have all reacted against the guidance, claiming that it locks researchers and clinicians into infl exible protocols that stifl e innovation and limit access to treatment. They say that laboratory-developed tests (LDTs) that individual institutions create and administer can adequately determine patient eligibility for personalized care. Vemurafenib ’ s creators at Plexxikon, a Berkeley, Calif. – based biotechnology company, rank among the guidance ’ s most outspoken critics; they claim it imposes excessive burdens on early-stage drug development. However, Plexxikon also anticipated diagnostic scrutiny from the FDA even before their BRAF inhibitor, then known as PLX-4032, had shown human proof-of-concept antitumor effects. So in 2005, to accelerate the drug ’ s development, Plexxikon teamed with Roche Molecular Diagnostics (RMD), in Pleasanton, Calif., on the development of an error-free system for detecting BRAF mutations. Producing the test took several years — partly because RMD had to devise a way to work with paraffi n-embedded, formalin-fi xed tumor samples, in which DNA is often damaged. From these efforts emerged the Cobas 4800 V600 Mutation Assay, a closed, PCRbased system that fl ags V600 → E mutations with a “yes” or “no” fl uorescence reaction. Plexxikon credits the test with propelling vemurafenib from discovery to approval in just 5 years, specifi cally by reducing the size of the trial population to V600 → E mutation carriers only. Vemurafenib ’ s label now indicates that because it was tested only in patients selected with the Cobas assay, doctors should avoid using it in BRAF-mutated patients selected using some other method. FDA spokesperson Erica Jefferson emphasized that the label doesn ’ t constitute regulation. “Product labels are intended to guide treatment decisions and provide clinical information,” she said. “The language is worded to allow for future approved tests for use in selecting patients for treatment.”