In the short time since the introduction of imatinib in 2001, investigators have already defined the primary mechanisms of drug resistance: mutations in the BCR-ABL gene that affect drug binding or an overall increase in the level of BCR-ABL protein due to gene amplification. In this issue, Branford and colleagues (page 276) report the first comprehensive mutation analysis of an unselected population of chronic myeloid leukemia (CML) patients treated with imatinib. The key finding is a very tight correlation between detection of a mutation and relapse, dispelling any remaining doubt about the causal role that mutations play in imatinib resistance.

A few additional points are worth noting. First, the authors find that mutations in the adenosine triphosphate (ATP)—binding loop confer a worse prognosis than other mutations, raising the possibility that early detection would mandate a change in treatment. Second, new clinical mutations are described here that were also picked up in a cleverly designed in vitro screen for imatinib resistance (Azam et al, Cell. 2003;112:831-843). Finally, the probability for finding a mutation increases with disease duration.

This last point is particularly important because it provides the first epidemiologic support for a clonal expansion model of imatinib resistance arising from pre-existing mutant subclones (Shah et al, Cancer Cell. 2002;2:117-125). The theory goes as follows. The CML clone makes sequence errors during DNA replication, some of which affect BCR-ABL. Over time, increasing clonal diversity raises the likelihood of generating imatinib-resistant subclones that expand in the setting of imatinib treatment. This model is based on data from patients who were treated with imatinib several years after their initial CML diagnosis. Will early imatinib treatment of newly diagnosed CML patients prevent this clonal evolution? Or is it too late? Time will tell.