On the Failure of Lung Cancer Drug Onartuzumab in a Phase III Clinical Trial

Most new cancer drugs fail clinical testing. Because they don’t make it to the pharmacy, we usually hear very little about them. But widespread media coverage made it hard to ignore the recent termination of a trial testing the drug onartuzumab. Details of the story raise concerns about the patient enrollment processes of some clinical trials.

Onartuzumab is a cancer drug being developed by the company Genentech/Roche. It is a targeted therapy that attacks the protein MET, which is sometimes found at high levels in patients with non-small lung cancer (NSCLC). High levels of MET are associated with poor prognosis and, importantly, resistance to drugs known as EGFR inhibitors. Erlotinib, an EGFR inhibitor, is currently a standard treatment for patients whose NSCLC tumors have EGFR mutations (as detected through molecular testing), because it was shown to be superior to chemotherapy in several pivotal clinical trials. Although erlotinib improves survival in this group of patients, many of them eventually develop resistance to erlotinib and stop responding. MET, when expressed at high levels, is known to be involved in the development of resistance to erlotinib.

To address the erlotinib resistance problem, onartuzumab was tested in patients with EGFR-mutant NSCLC. A phase II clinical trial compared treatment with a combination of onartuzumab and erlotinib, versus erlotinib alone. The results were published in the Journal of Clinical Oncology late last year. The onartuzumab/erlotinib combo did not show an overall survival advantage for the patients.

These results are not too surprising when the design of the trial is taken into account. Even though onartuzumab targets MET and erlotinib resistance is known to be associated with high levels of MET (‘MET-positive’), the researchers enrolled many patients without elevated MET (‘MET-negative’). Why were all patients, rather than only MET-positive patients enrolled in the trial? Why use a drug targeting MET in patients whose tumors do not have high levels of MET? The practice, still ongoing, of enrolling ‘all comers’ into trials of targeted therapies without testing for relevant genetic mutations or other molecular alterations, has been criticized and deplored numerous times, including by me.

However, in their publication the study authors reported further analysis of the responses and came to the conclusion that, when MET levels were taken into account, the onartuzumab/erlotinib combo showed a clear advantage in patients with MET-positive tumors. The MET-positive patients who received both drugs had an overall survival time of 12.6 months compared to 3.8 months for MET-positive patients who received erlotinib only. Strikingly, patients with MET-negative tumors fared a lot worse if they received both drugs rather than erlotinib alone. As the authors of the study concluded, “These results combined with the worse outcomes observed in MET-negative patients treated with onartuzumab highlight the importance of diagnostic testing in drug development.” Indeed.

With this lesson learned, the phase III trial enrolled only patients with MET-positive tumors, and, based on the results of the phase II trial, expectations were high. However, the phase III study was stopped in early March due to futility, that is, there was no evidence that the addition of onartuzumab to erlotinib has any positive effect.

How to explain the lack of onartuzumab efficacy that was so evident in the phase II trial? Was there a problem with the analysis of the results of the phase II trial? Was there a problem with the test for MET levels in patients’ tumors?

In any case, Genentech/Roche is in all likelihood scrambling now to explain the glaring difference between the results of the phase II and III trials. Moreover, onartuzumab is currently being tested in 14 other trials, some still recruiting patients and others filled, for different malignancies. It looks like the majority of the onartuzumab trials are not exclusively enrolling patients whose tumors are MET-positive—the poor practice continues. If the observation that MET-negative patients with NSCLC did much worse when treated with onartuzumab holds true, it could have serious consequences for patients with various other MET-negative cancers receiving the drug in the ongoing onartuzumab trials.