Tyrosine kinases may be constitutively activated in leukemias as fusion genes by intragenic juxtamembrane deletion or in frame insertions, or by “activation loop” mutations in the catalytic domain. In this issue, Grundler and colleagues (page 646) demonstrate that there is varying sensitivity of 3 activation loop mutations in FLT3 (Asp835Try, Ile836del, and Ile836Met+Arg) to the tyrosine kinase inhibitors AG1296, PKC412, and SU5614, respectively. Most small molecule tyrosine kinase inhibitors are ATP analogs that bind with high affinity to the ATP binding site and preclude access of ATP and substrate. It might therefore be predicted that certain activation loop mutations would confer relative resistance to small molecule inhibitors. Indeed, Shah et al have shown that a collection of mutations in BCR-ABL may not only enhance kinase activation and confer a proliferative advantage but may simultaneously confer primary resistance to imatinib (Shah et al, Cancer Cell. 2002;2:117-125).

What are the implications of these findings? From a structural perspective, the observation of variable sensitivity of activation loop mutations to small molecules of known structure should provide valuable insights into the structure-function relationships of tyrosine kinases.

There are also important clinical implications of these findings with regard to inclusion criteria for trials of FLT3 inhibitors in acute myeloid leukemia (AML). Most phase 2 trial designs have required genotypic evidence of either juxtamembrane or activation loop mutations for inclusion in clinical trials. The authors tested sensitivity of only 3 of the dozen or so reported FLT3 activation loop mutations. But it seems clear from their data that each of the known FLT3 activation loop mutations should be tested for sensitivity to any tyrosine kinase inhibitor being considered for clinical trials, and that inclusion criteria should reflect sensitivity of the activation loop mutation to the given FLT3 inhibitor. For example, it may not be appropriate to treat AML patients with FLT3 activation loop mutations that are known to be resistant to a given inhibitor. In this context it should also be noted that rare patients have both juxtamembrane and activation loop mutations on the same FLT3 allele, indicating that all patients should have sequence analysis of both domains prior to inclusion in clinical trials.

These considerations add complexity to clinical trial design, including the apparent need to stratify patients based on the genotype of the FLT3 juxtamembrane domain and the activation loop.