ROCKVILLE, Md. (
) --I'm going to dissect the results from
phase II study of its experimental anti-depressant Serdaxin that were announced last week.
, but the market said otherwise. Rexahn lost about 50% of its value Tuesday, falling to $1.76 a share, following the release of the Serdaxin data on Tuesday afternoon. But the stock recovered a bit after the company tried to further explain the study results Wednesday. By Friday, the stock closed at $2.84 a share.
I'm often asked how I go about evaluating clinical data. Here's how I do it. Since all we have is Rexahn's press release describing the Serdaxin phase II study in depression, that's what I'll use to parse the data:
Rexahn's press release begins:
"The randomized, double blind, placebo controlled and dose ranging clinical trial enrolled 77 patients at multiple sites in the United States. The results of the subgroup analysis are compelling and warrant further study in a larger phase 2 trial."
A good start. Rexahn designed the phase II study with a placebo control arm, to which we can compare and evaluate Serdaxin's efficacy and safety. The study is dose ranging, too, so we can see whether escalating doses of Serdaxin improve efficacy and/or cause added side effects.
"The results of a subgroup analysis..." The first warning bell goes off and we've barely gotten started. When a company highlights a positive subgroup analysis, it often means the overall study was negative, forcing the company to go on a fishing expedition through the data to find a group of patients with positive results. Let's see what Rexahn says next about the Serdaxin data.
"In the subgroup analysis, the study showed that patients with severe MDD taking 5 mg of Serdaxin had significant improvement in Montgomery-Asberg Depression Rating Scale (MADRS) scores after 8 weeks of treatment, compared to placebo. Among the 77 patients, 53 patients were classified as having severe MDD. Of the 14 patients treated with 5 mg of Serdaxin MADRS scores improved by 55.6%, compared to only 34.0% in the placebo group (n=14), which was statistically significant (p less than 0.041) on an intent to treat basis."
Alright, here's the positive data referenced in the first paragraph. The subgroup is patients with severe depression. Treatment with 5 mg of Serdaxin resulted in an almost 56% improvement in MADRS scores compared to a 34% improvement for patients treated with placebo. The result was statistically significant, although just barely, with a p value of less than 0.041. The MADRS scale is a rigorous and commonly used efficacy measure in depression studies, so no quibbles there.
But Rexahn's disclosure of the "positive" data in this paragraph comes with some gaping discrepancies. Let's start with the identified subgroup of patients with severe depression, which accounts for 69% (53 of 77) of the patients enrolled. Was this a prospectively defined subgroup of patients? Rexahn doesn't say, but I have my doubts.
If Rexahn had meant to study Serdaxin in patients with mild, moderate and severe depression, wouldn't the company have enrolled more patients in the former two categories? By inference, patients with mild and moderate depression only make up 31% of the study. And since Rexahn doesn't tell us anything about Serdaxin's performance in mild and moderate depressed patients I'm assuming the drug had no positive effect there.
Rexahn also fails to disclose anything about the overall results of the study, which is odd.
The red flags identified in the first paragraph are multiplying. This subgroup analysis appears to be a data-mining expedition. Rexahn doesn't tell us anything about how patients with severe depression were defined clinically, which introduces bias, especially if this subgroup was identified after the company had all the Serdaxin data in hand.
More trouble. Rexahn appears to be subgrouping its subgroup. Notice that only 14 patients labeled with severe depression were treated with 5 mg of Serdaxin. But Rexahn told us that 53 patients in the study had severe depression. What happened to these missing patients?
I assume some of the missing severely depressed patients were assigned to the 14-patient placebo arm, but the others?
Let's go back to the first paragraph, which told us that the study was dose ranging. That means some patient with severe depression -- the "positive subgroup" -- must have been treated with higher or lower doses of Serdaxin.
Let's leave the Rexahn press release for a moment to see if we can find out anything about the other Serdaxin doses tested in this phase II study. Sure enough, the
. Aha! Rexahn also assigned patients to receive treatment with
Rexahn tells us that 5 mg dose of Serdaxin worked, but doesn't say anything about the two higher doses. That's not an encouraging sign. After all, if patients responded to the lowest 5 mg dose, why wouldn't they also respond, perhaps even better, to higher doses? That's called a dose response and is fairly common - and a positive indicator of efficacy -- in studies such as this.
So, in two paragraphs, we've gone from "compelling" evidence of Serdaxin's efficacy in depression to a retrospective analysis consisting of a subgroup (the low, 5 mg Serdaxin dose) of a subgroup (severely depressed patients .)
Will the Serdaxin data improved as we move to the next paragraph in Rexahn's press release?
"In addition, 64.3% of patients with severe MDD treated with the 5 mg of Serdaxin were considered Responders compared to 28.6% in the placebo group (p less than 0.0581). A Responder is a patient with a change from baseline in MADRS score of greater than or equal to 50% after treatment. Additionally, 42.9% of patients in the treatment group were in remission with a MADRS score of less than or equal to 12 after treatment, at 8 weeks versus 14.3% in the placebo arm (p less than 0.209)."
Rexahn is providing us with additional efficacy analyses, presumably secondary endpoints, to bolster the case for Serdaxin. That's fine, except again, the only data we're given is from patients with severe depression treated with 5 mg of Serdaxin. It's that darn subgroup of a subgroup again.
And those p values tied to the secondary endpoints aren't pretty, neither reaches the level of statistical significance. In fact, the latter, measuring remission rates for patients in the study, doesn't even come close.
Rexahn doesn't say so explicitly, but I'll assume the argument it's making is that the efficacy trends in both secondary endpoints strongly favor Serdaxin and that the study enrolled too few patients to be powered for statistical significance.
Fine, but it's just as legitimate to look at these data and conclude Serdaxin failed on two of three different measures of efficacy in the company's chosen subgroup of patients with severe depression.
More troubling signs from the Serdaxin data can be found if we revisit the details of the
The first bullet point under "Secondary Outcomes Measures" is "Changes from baseline on the Ham-D."
The Ham-D is another widely used measure of efficacy in depression trials. Rexahn lists analysis by Ham-D as the first of the secondary endpoints in the study, yet that analysis is missing entirely from the company's press release. Likewise, Rexanh tells us nothing about some of the other secondary efficacy endpoints that were supposed to be analyzed in the Serdaxin study, according to the details of the study listed on ClinicalTrials.gov.
If depression patients were truly responding to Serdaxin, wouldn't that response be picked up by more than one efficacy scale? Why did Rexahn selectively disclose a tiny amount of the data from this study? It's not like Rexahn didn't have time to conduct a full analysis, the company first announced top-line results from the
Six months later, Rexahn has still not published data from the study, nor has it been presented at a medical meeting. And the data Rexahn did choose to disclose, via press release, is filled with holes and raises more questions than it answers.
More from the Rexahn press release:
"The trial also validates the earlier results which demonstrated Serdaxin to be safe and well tolerated without the appearance of serious side effects that are commonly linked to currently marketed antidepressant drugs, such as selective serotonin uptake inhibitors (SSRI), serotonin-norepinephrine reuptake inhibitors (SNRI), and tricyclic antidepressants (TCA). Even though the overall study did not achieve statistical significance, we believe, based on the statistically significant subgroup results, Rexahn plans to commence a Phase IIb clinical trial in the second half this year."
Ah, so here's where Rexahn tells investors that the study failed overall. Now, I'm actually willing to cut Rexahn a little slack here. A study enrolling 77 patients across four treatment arms is probably not powered strongly enough to achieve statistical significance. However, that doesn't give the company license to data-mine the study and declare victory because some arbitrarily defined subgroup of a subgroup appears to have responded to the drug.
Moreover, Rexahn does a lousy job of disclosing the side effect profile of Serdaxin. All drugs have side effects, yet Rexahn tells us nothing except that patients didn't experience side effects commonly seen with other currently marketed anti-depressants.
But what about side effects commonly seen in patients taking
, which is the active ingredient in Serdaxin? Clavulanic acid is an old drug and a component of the antibiotic Augmentin, used to overcome resistance to certain strains of bacteria.
Rexahn chooses not to include any of this information in its press release either, but that's what Google is for. Use it.
I wonder if the millions of people treated with Augmentin over the years have ever reported feeling happier and/or less depressed? Rexahn has some animal data showing anti-depressant attributes of clavulanic acid (Serdaxin) but judging by the phase II study in humans, the company has a lot of work to do before it can make the case for developing an old-school antibiotic resistance fighter into a new, cutting edge anti-depressant.
Towards the end of Rexahn's press release, we get this information:
"Rexahn is currently in discussions with several major pharmaceutical companies with the goal of identifying a potential strategic partner to assist in the development and commercialization of Serdaxin. However, there can be no assurances that these discussions would result in a commercial arrangement."
Says who? Rexahn? Statements like this are one sided and meaningless. The suggestion that a large drug company is going to risk tens of millions of dollars in drug developing monies, perhaps even more, based on results culled from a retrospective analysis of 14 patients seems more than a bit far fetched.
Rexahn might be having discussions with several major pharmaceuticals firms, but the conversation is probably a bit one-sided and might sound something like this:
"Thanks Rexahn, but call us back when you have more data."
When I last wrote about
, I rapped chief executive Chang Ahn on the knuckles gently for spending too much time promoting his stock and too little effort on qualit, drug development. In particular, I urged Ahn to be more transparent with his clinical data.
Judging by last week's press release, Ahn still isn't listening to good advice offered.
On a related note: Investors interested in strong phase II data in depression should take a look at
and its drug
-- Reported by Adam Feuerstein in Boston.
Follow Adam Feuerstein on
Adam Feuerstein writes regularly for TheStreet.com. In keeping with TSC's editorial policy, he doesn't own or short individual stocks, although he owns stock in TheStreet.com. He also doesn't invest in hedge funds or other private investment partnerships. Feuerstein appreciates your feedback;
to send him an email.