Testimony at FDA Advisory Committee Oncology Meeting on Mylotarg (“GO”)

Jack Mitchell, National Center for Health Research, July 12, 2017

Thank you for the opportunity to speak today.  I’m Jack Mitchell, Director of Health Policy at the National Center for Health Research (NCHR).  Our research center analyzes medical and scientific data to provide objective health information to patients, providers and policy makers.  We do not accept funding from pharmaceutical or medical device companies, and so I have no conflicts of interest to report.

I’m not a scientist or clinician, but I previously worked in a senior position in FDA’s Office of the Commissioner and we have a number of science and public health PhDs on our staff.  I’m presenting the organization’s views on behalf of the many patients and consumers whom we represent.

While we strongly support the need for more and better treatments for AML (Acute Myeloid Leukemia) patients, we are very concerned about the data used to support the application for GO.  Problems with the trial design raise questions about its validity.

First of all, the only pivotal trial was open label, which increases the risk for bias.  The purpose of blinding in a clinical trial is to control for the placebo effect, since the knowledge that one is taking the newest experimental drug tends to encourage patients and clinicians to have a greater belief in a perceived effectiveness.

Second, all lower grade safety events and some important severe safety events were collected retrospectively, which increases the risk for inaccuracies.  And, which as FDA presenters have noted, limits the analysis of the safety profile.

Third, the trial took place only in France.  This is of note because there are numerous examples of medical products that do not work as well in American patients as they do in patients in other countries.  French patients can have different health habits and health care.  These issues would raise concerns, even if the data supporting approval was strong, which we believe they are not.  Instead, it is not clear that the data support the efficacy or safety of GO.

As I noted previously, the application is based on a single pivotal trial along with a review of the literature.  The pivotal trial does not provide evidence for overall survival, and the previous clinical trial included in the literature review found an inconsistent effect of GO in overall survival.  It is important to remember that this drug was approved based on a single trial and later removed from the market, in part, because the post-market study did not demonstrate effectiveness.

The pivotal trial and literature review do demonstrate improvement in event-free survival.  The trial also shows an improvement in relapse-free survival.  However, the important metric is overall survival, which is not clearly demonstrated.  FDA reviewers and the sponsor’s scientists showed that event-free survival does not correlate well with overall survival.  This especially is a problem in an open label study where the placebo effect can’t be controlled.

Our research center recently published an article in an AMA journal revealing that many cancer drugs have been approved based on surrogate endpoints, such as event-free survival.  But later studies have found that those drugs did not improve overall survival or quality of life.  We found that patients and their insurers were spending $100,000 or more and suffering serious adverse events for treatments that often had no measurable benefit for their health or survival.

The change in dosing does appear to reduce adverse events compared to the earlier version of this medication.  Nevertheless, there were still serious adverse events that can result in death.  The drug was associated with increased bleeding events, including four fatal hemorrhages, and liver disorders, including three fatal cases of VOD (veno occlusive disease).  There were no fatal hemorrhage or VOD events that occurred without exposure to the drug.   However, we acknowledge the sponsor’s efforts to address the ongoing VOD issue and risk profile.

It’s noteworthy that these results were in a clinical trial where patients are carefully monitored.  Patients in the real world are typically monitored less carefully than patients in clinical trials.  As a result, it is possible that more patients could continue on a drug causing serious adverse events because they hope the drug will improve their condition, putting themselves at increased risk.  Well-intentioned doctors who are unaware of the history of the drug may also decide to increase the dose of patients who are not improving, putting patients at greater risk for adverse events without improving the chance for survival.

In summary, surrogate endpoints, such as event-free survival, often do not predict overall survival or other measures of improved health or quality of life.  Given the research design, only one pivotal study, the lack of U.S. patients, and a less than convincing literature review, the data do not support sufficient support for approval.  The studies, in our view, do not provide strong evidence that GO is effective and there are still continuing safety concerns.

We believe that the evidence does not indicate that the benefits outweigh the risks, which is the most single important consideration before you today.  Thank you for providing this opportunity to present our views.