In the past two weeks, the US Food and Drug Administration has approved two targeted cancer drugs -- and companion diagnostics to go with them. Not only do the approvals provide new therapies for patients, but some investigators think they show a positive trend in drug development.
On August 17, the agency approved vemurafenib (Zelboraf) for the treatment of late-stage or unresectable melanoma in patients whose tumors carry the BRAF V600E mutation.
On August 26, the agency approved crizotinib (Xalkori) for patients with locally advanced or metastatic non-small cell lung cancers (NSCLC) whose tumors carry a mutation in the anaplastic lymphoma kinase (ALK) gene.
“I believe that these are very important approvals,” said David R. Gandara, MD, Professor of Medicine and Associate Director of Clinical Research at the University of California Davis Cancer Center in Sacramento.
“Both because the drugs will benefit patients, but also because they reflect an increasing paradigm shift towards personalized medicine through predictive biomarkers based on individual patient tumor genetics. So the theme here is incredibly important.”
George W. Sledge Jr., MD, Immediate Past President of ASCO and the Ballve-Lantero Professor of Oncology at the Indiana University Melvin and Bren Simon Cancer Center (and an OT columnist/blogger), sees the approvals and the companion diagnostics as part of a larger trend that started with trastuzumab, the use of which was restricted to patients with HER2-positive tumors.
“While it is possible to imagine vemurafenib without a companion diagnostic (the mutation occurs in half of melanoma), it is inconceivable that crizotinib could be used or approved without the companion diagnostic, given that ALK+ patients constitute only four to five percent of NSCLC,” he said.
Vemurafenib Improves Survival
The agency approved vemurafenib on the basis of an international randomized trial that compared the experimental drug with dacarabazine in 675 patients. All patients were required to have the V600E mutation. The estimated median survival for the patients treated with vemurafenib has not been reached (with 74% still alive), compared with eight months for patients treated with dacarabazine (with 64% still alive), according to the agency.
The drug is the second agent approved for patients with advanced melanoma in less than a year, which is remarkable. (Ipilimumab was approved in March.) And the agency approved the drug well ahead of the October 28th goal.
Lack of Survival Data with Crizotinib
Crizotinib also received an earlier than expected approval, with the decision expected for late September. The agency gave the drug an accelerated approval based on two single-arm phase II studies that included 255 patients. In one study, the objective response rate was 50% and the median duration of response was 42 weeks. In the second study, the response rate was 61% and the median duration of response was 48 weeks.
Neither study assessed the drug’s impact on overall survival -- which I think is important -- though the response rate was high and patients reported feeling better.
As part of the approval process, Pfizer has agreed to continue two phase III trials. The first is an open-label randomized trial comparing crizotinib with pemetrexed or docetaxel as second-line therapy. The primary endpoint is progression-free survival and patients will be allowed to cross over on progression.
While that is clearly the ethical design, given the lethality of the disease, allowing crossover means that it will be difficult, if not impossible, to assess the drug’s impact on overall survival.
A second phase III trial comparing crizotinib with pemetrexed-platinum combination therapy in previously untreated patients may have a better chance to reveal overall survival benefit, with landmark survival analyses at 6 and 12 months.
I wonder if an unintentional by-product of the trend for biomarker-driven therapies, such as these two agents, will be the demise of overall survival as a primary endpoint. After all, if a novel targeted drug shows dramatic results in a select group of patients, it may be hard to justify randomizing patients to another (presumably less effective) agent without crossover.
Maybe the need for new endpoints, or a greater reliance on progression-free survival and other surrogate endpoints, will be part of the cost of success for the targeted therapy field.
The High Financial Cost of Success
That isn’t likely to be the only cost, however. The price tag for vemurafenib is $9,400 per month, and for crizotinib, it is $9,600 per month.
That means that a course of vemurafenib therapy will be about $50,000 if a patient has a progression-free survival similar to the median of 5.3 months seen in the trial. (Ipilimumab, by comparison, costs $120,000 for the standard four-dose course.)
If a patient on crizotinib has a median response of 48 weeks, the price tag would be about $115,000.
Are the prices too high? I don’t know, but the numbers are enough to make many in the field worry about how much the system can bear.
Dr. Gandara, though, is optimistic. “Although new drugs are always expensive, this sort of biomarker-driven targeted therapy is actually cost effective, by finding the right treatment for the right patient,” he said.
(For a more extensive discussion of drug costs, you might want to read Dr. Len’s blog from the American Cancer Society.)
The good news is that the BRAF diagnostic test is expected to cost between $120 and $150, which seems very reasonable. And it is low enough that it won’t be a burden on patients who find they are not eligible for the drug. (I haven’t seen any estimates for the cost of ALK companion diagnostic yet.)
Cost issues aside, though, the approvals are surely good news for the cancer community -- and show the power of co-developing a drug and diagnostic test.
In response to my post yesterday, a Pfizer spokesperson noted that the researchers presented crizotinib overall survival data from the single-arm phase II trials at ASCO. Robert Carlson covered that presentation for OT and wrote the following:
"ALK-positive patients treated with crizotinib had a one-year overall survival rate of 74% and a two-year overall survival rate of 54%. Median overall survival had not been reached. This compared with outcomes for the ALK-positive controls of one-year overall survival of 73% and two-year overall survival of 33%. Median overall survival was 20 months."
My point, however, is that overall survival data from randomized trials will be difficult if not impossible to obtain when crossover is allowed (as it ethically should be). Comparisons between groups in non-randomized trials – such as the single-arm phase II crizotinib studies on which the accelerated approval was based and from which the overall survival numbers come – are fraught with problems and are not statistically valid.
So while the company and the academic researchers did collect and report the overall survival data, I think we need to view it with more than a grain of salt. Not because there is anything amiss, but rather because you simply can't make that comparison in a non-randomized setting.
Therefore, I'm back to what I said before. I think one cost of success with targeted agents may be the need to rely more on surrogate markers in randomized trials.
I'd be very interested to hear other people's thoughts on this issue.