Greetings,
There was once a time in history when doctors prescribed blood-letting, and were convinced that it was effective – and could cite many success stories. In comparison, homeopathy (water with trace amounts of different compounds) along with good hygiene appeared effective!
Each medical practice (disproven by modern scientific methods) was based on a prevailing theory and substantial clinical observation.
There are many such examples of theories that when tested rigorously just didn’t hold up, (such as HRT for women) and yet, clinical observations seems the basis for a good deal of professional opposition (about 50/50) on the recent Advisory committee’s recommendation (12-1) against approval of Avastin for metastatic breast cancer.
(See Nine Members of the Oncology Community Speak Out about Bevacizumab’s Role in Metastatic Breast Cancer http://bit.ly/dxbLUp
So how do you measure clinical benefit, if not by theory (an infinite #) and observation, which is influenced mightily by what we expect and want to see, and our financial interests if any?
For example, and this could seem harsh, how many professional opinions on Avastin are explained by a natural reluctance to admit to prescribing an ineffective therapy?
Some background:
Today a treatment is proven to provide clinical benefit if it improves survival compared to the natural course of the disease, or the disease treated differently, but also if it improves quality of life (a separate topic).
… But since we can’t know if any individual would have done better with therapy X compared to Y, the testing must be done on large groups, randomly assigned to one arm of the study or the other. If one group does better than the other (lives longer or better), we have evidence of clinical benefit that is firm and objective.
So the randomized controlled trial (RCT) design is a triumph of modern clinical science as it protects us from our own biases.
However, WHAT and WHEN you measure the effects of therapy can make even the results of a RCT controversial (as in the Avastin example above), a drug, which was conditionally approved based on measured improvement in progression free survival (PFS) for those receiving Avastin in a prior study.
So what is PFS and why is it used instead of overall survival?
PFS generally measures time to progression, relapse, and to death from ANY cause.
The rationale for including “death from any cause” – even if not apparently related to the disease or treatment – is that it’s not always possible to know what’s related; and accidental or unrelated causes will balance out when comparing groups treated with different protocols.
In FDA deliberations, PFS is also weighed against offsetting toxicities. PFS is a surrogate for survival because survival benefit can take too long to measure, particularly for the indolent cancers that have long survival; and because the measure is confounded when patients use additional therapies.
It appears that in the follow-up Avastin studies, done to confirm the finding of the first, the magnitude of PFS for Avastin was smaller and the toxicities were found to be more significant. Further, Avastin did not provide a survival advantage at the point it was measured.
What will be FDA’s ruling?
I expect the agency will remind that an improvement in PFS must be sufficient to reasonably predict survival benefit. If the PFS advantage is small and the price for that benefit is high in respect to the toxicities, I expect it will go along with the Advisory committee to deny full approval,
…. but regarding conditional approval, it may simply require additional study. If so, the task for the sponsor will be to figure out which subgroup of patients may benefit from this drug (find a biomarker) … which is what patients want, and need!
For patients, in effect, the conditional approval remains, because the drug (approved for other indications) can be used off-label and also can be tested further.
Consider also that for this indication there are other effective protocols, such as the chemo induction therapy used as the control. So arguments supporting approval - such as that the small improvements in PFS or response rate have been used as the basis for other approvals – I don’t find persuasive.
Actually, I started this piece because I wanted to write about PFS as a composite measure of benefit for indolent lymphomas. (A composite because it measures time to progression AND mortality.)
It is clearly a necessary metric for this indication, but I wonder if its meaning is inflated in studies that have low and/or equivalent mortality in the study arms? In such cases, we could more accurately call it Time to Progression so we don’t assign to it a meaning (the survival part) it doesn’t deserve.
Anyhow, in summary:
Theory and even professional observation are each unreliable as evidence. The theory-holder is especially vulnerable to selective observation (seeing only what fits his theory); the theory-inventor and commercial sponsor even more so!
Opinion by Karl Schwartz
President, Patients Against Lymphoma