Thread by @THilalMD

Our paper with @VPrasadMDMPH and @sonbol_bassam JUST OUT in @JAMAOnc

How often are cancer drugs approved based on trials that compared them to suboptimal controls?

Quick thread jamanetwork.com/journals/jamao…

First, this is a tricky question to answer. The choice of what is considered “standard” therapy is somewhat subjective. Here is what it’s NOT:

-> It is not a drug that was proven to be superior to placebo based on a phase 3 RCT (in fact, many are not)

SOC is determined by practicing oncologists and national guidelines. Some are based on single-arm phase 2 trials, some are randomized phase 2 vs placebo.

Any drug that is likely more effective than placebo can be considered SOC (many established cancer therapies)

If the only bar for SOC is a phase 3 RCT demonstrating OS benefit, then all these approvals would’ve been compared to placebo.

Many SOC therapies did not become SOC because of phase 3 data showing OS benefit.

Alright, now that SOC is defined.

Shouldn’t any new drug, moving forward, be able to demonstrate effectiveness above SOC?

Now that there is a selection of active agents to build upon, shouldn’t we want to improve SOC?

Enter our study.

We looked at all the FDA approvals from this website between January 2013 through July 2018 (fda.gov/drugs/resource…)

We excluded all approvals based on single-arm trials (not a small amount).

We then evaluated the control arms and start date of participant accrual on trials.

More on this in this @statnews piece
statnews.com/2019/05/02/sub…

It takes time for trial protocols to be written and go through IRBs, and other bureaucratic nonsense, so we have a 1 year “grace period” during which change in SOC may occur but control arm choice cannot be changed to reflect that.

We then looked at national guidelines (NCCN) and review articles that were written 1 year prior to enrollment on trials to see what was considered SOC by the community at that time.

When control arm was limited and did not include SOC or omitted an agent altogether, it was considered suboptimal.

This piece in @MedscapeOnc expands on how the categories of suboptimal control medscape.com/viewarticle/91…

There are nuances to selecting control arm (e.g. international trials where control arm must be available in other countries)

Listen to my discussion with @VPrasadMDMPH on @Plenary_Session podcasts.apple.com/us/podcast/ple…

The end result - 17%.

17% of RCTs that led to cancer drug approvals between 2013 and 2018 were based on control arms deemed suboptimal.

We can argue about a few of the selections, and you may not agree with all 17%, but that’s ok.

I hope you will see that this phenomenon occurs more often than necessary, whether it’s 17%, 15%, or 12%.

If there is one thing I want to highlight it’s this -

Oncologists need to examine the evidence for themselves when deciding on using a new approved agent and ask themselves - is this drug truly better than what is already available? And by how much? And at what cost?

There will be difficulties in insuring that control arms are always optimal and reflective of SOC when trials are conducted internationally. But that’s a question we have to wrestle with.

That’s it!

Please share this thread. And if you have thoughts or suggestions about ways to improve things moving forward, please reply below.