GW Pharmaceuticals (GWPH) ended up 7.6% yesterday after announcing in a press release headline that the company had "Positive Proof of Concept Data in Schizophrenia" for its cannabidiol, or CBD, drug.
Sounds great, right? With proof of concept in hand, the company just needs to confirm the phase 2 results and everything is grand. Hooray for science. And medical marijuana derivatives. And patients. And investors.
There's just one problem: proof of concept implies that you had a hypothesis, tested it, and it turned out to work, albeit on a scale smaller than is necessary to get the drug approved.
But GW Pharmaceuticals doesn't have hypothesis-driven data. Instead, the biotech says that "over a series of exploratory endpoints, CBD was consistently superior to placebo." Seven paragraphs into the press release the company states, "As a phase 2a proof of concept study, there was no single primary endpoint, but a series of exploratory endpoints."
Now that doesn't exactly jive with the original clinicaltrials.gov entry for the trial, which lists a primary outcome for the study as the "change from baseline to the end of treatment in Positive and Negative Syndrome Scale (PANSS) total score" and then lists 28 secondary endpoints.
Yesterday the company changed the cinicaltrials.gov entry so that it now reads that there are 13 primary outcome measures and 16 secondary outcomes. As you hopefully learned while doing your sixth grade science fair project, changing the hypothesis after the data has been presented is not how the scientific method is supposed to work.
You won't find the word "exploratory" anywhere on either listing.
The term "exploratory endpoint" is typically used when a primary endpoint fails; then all the secondary endpoints become exploratory because the hypothesis failed. That may be what has happened here, but GW Pharmaceuticals certainly isn't fessing up to it.
Admittedly, I've never seen a clinicaltrials.gov listing that didn't have a primary endpoint (as it should be since the goal of a clinical trial should be to test a hypothesis). So if GW Pharmaceuticals really wanted to do a trial with just exploratory endpoints, maybe it was still required to list a primary endpoint (or 13).
But the problem is that GW Pharmaceuticals can't say every endpoint is exploratory and then claim success when some of the endpoints come up positive. If you test enough endpoints, you'll eventually find something that shows a difference between the drug and placebo because of chance alone.
The press release only mentions three endpoints where CBD was superior to placebo with a p-value of less than 0.05, meaning there's less than a 5% chance that it happened by chance alone. None of the three were the aforementioned change from baseline to the end of treatment in the PANSS total score that was originally listed as the primary endpoint.
Rather than conclude that this data shows positive proof of concept, it's more accurate to say this data gives GW Pharmaceuticals a hypothesis to test in an additional trial. The company needs to set one of the positive endpoints as the primary endpoint and see whether CBD is superior to placebo in that trial.
In addition to the data mining to find something positive to say, investors should be a little worried because schizophrenia, by its nature, has symptoms that can change. It's possible that the positive endpoints are just an artifact caused by the variability of schizophrenia.
Another potential for variability in the data comes from where the clinical trial was run. A vast majority of the investigators were in Poland and Romania with just three in the United Kingdom. While everything may be fine, when you move clinical trials into countries with less developed medical systems, there's potential for data to be reported improperly.
The good news for investors is that it doesn't appear that GW Pharmaceuticals is ready to pull the trigger and continue to push CBD into another schizophrenia trial. "Similar to our approach for Epidiolex, we believe that our future research in this area may lie within pediatric orphan neuropsychiatric indications and we intend to explore this as a focus for future trials," Justin Gover, GW Pharmaceuticals' CEO said in the press release.
But without another phase 2 trial to confirm the initial findings, it's hard to see how GW Pharmaceuticals is worth more than it was before exploratory endpoints came back positive.