Imatinib mesylate (Glivec) potently inhibits the kinase activity of the oncogenic fusion protein BCR–ABL and is an effective treatment for chronic myeloid leukaemia. Although imatinib seems to be well tolerated by most patients, more than 60% of patients on imatinib in clinical trials develop peripheral oedema, a possible sign of cardiotoxicity. However, heart function was not assessed in any of the clinical trials with this agent. Thomas Force and colleagues report that 10 patients who previously had normal heart function developed severe heart failure after the initiation of imatinib therapy. They found that imatinib is not directly toxic, but that the inhibition of one of its targets, ABL, triggers the stress response in cardiomyocytes and induces cell death.

Myocardial biopsy samples taken from patients showed prominent membrane whorls in the myocytes, an abnormality that is a characteristic of toxin-induced myopathies. The cells also showed pleiomorphic mitochondria and dilated endoplasmic reticulum (ER), which are signs of stress.

To investigate this further, the authors treated healthy mice with imatinib and saw similar structural changes in the mouse hearts. Doses of imatinib given to produce blood concentrations comparable to those in humans led to left ventricular dysfunction. Imatinib seemed to cause necrotic cell death (as seen by the dose-dependent collapse of mitochondrial membrane potential, release of cytochrome c and pronounced cytosolic vacuolization) rather than apoptosis. Gene transfer of an imatinib-resistant mutant of ABL inhibited the imatinib-induced release of cytochrome c and protected from cell death. This suggests that the inhibition of ABL by imatinib is the mechanism of cardiomyocyte toxicity.

What are the mechanisms that regulate imatinib-induced cardiomyocyte death? Because of the observed ER dilation, Force and colleagues investigated the ER stress-response pathway and found that imatinib activated both the EIF2α and IRE1 parts of this pathway in imatinib-treated mice. Although this response is initially protective, if the inducing stress to cells is not relieved — as occurs with prolonged imatinib therapy — IRE1 can signal cell death by activating the pro-death JNK pathway. The authors found that JNK was activated in the hearts of imatinib-treated mice and that this activation was reduced by treatment with salubrinal, a small-molecule inhibitor of EIF2α dephosphorylation. Furthermore, the inhibition of either EIF2α dephosphorylation with salubrinal or JNK activity with a peptide inhibitor rendered cardiomyocytes resistant to the imatinib-induced collapse of mitochondrial membrane potential and cell death. Therefore, the imatinib-mediated induction of the ER stress-response pathway leads to cell death through the activation of JNKs.

These findings not only suggest that patients who are on imatinib should be monitored closely for signs of left ventricular dysfunction, but also that clinical trials of new agents that target ABL should prospectively assess left ventricular function so that rates of cardiotoxicity can be determined.