Longboard Pharmaceuticals (LBPH): Watch the baseline
Originally published 1/3/2024 (on 1/2/24 the stock opened at $6 and closed at $25)
Sometimes the results of a positive clinical trial reflect the efficacy of the drug itself: For example, the new generation of GLP-1 receptor agonists have achieved their encouraging clinical endpoints in trials for the very reason that they work — at least for the stated goal of reducing a person’s body weight.
But sometimes the positive results derived from a clinical trial — particularly an early stage clinical trial — reflect the methodologies of the trial itself (i.e., the trial design, or the patient randomization amongst treatment groups), and not so much what the drug itself was doing.
When the results of a clinical trial are released, and the valuation of the company that sponsored the trial quadruples in the course of a few hours — as happened yesterday for Longboard Pharmaceuticals (LBPH) — any reasonable person might conclude that this company has a home run on its hands, and that the drug in question really works.
But that same reasonable person might not realize that the price was driven up by the actions of unreasonable people who didn’t bother to examine some fairly obvious differences between the treatment group and the control group (which, in the case of the LBPH study, was a group that received placebo). And, in this case, the results that were obtained in the trial may not reflect the efficacy of the candidate drug, but rather the differences between the patients in each treatment group.
So let’s get under the hood.
IMPACT OF A BASELINE IMBALANCE ON TRIAL RESULTS
LBPH sponsored a phase 1b/2a study of bexicaserin, a 5HTc agonist for treatment of refractory seizure disorders in patients with very serious developmental diseases associated with seizures, mental retardation and other complications. Patients with Dravet Syndrome and Lennox-Gastaut Syndrome (which were included in this study of patients with heterogenous diseases) are neurologically devastated at an early age and have shortened life expectancies.
And I think it would be fair to say that the results of the trial may have achieved a proof of concept, as there was a 50% decrease in seizure rates in the treatment group. But there are significant problems with the study design, and even a few different ways of interpreting the results, that should caution anyone from concluding a proof of effect.
The first problem is the significant baseline imbalance between the experimental and control groups. For example, in the treatment group, at baseline these patients were having almost 40 seizures a day — whereas in the placebo group, at baseline, they were only having about 20 seizures a day. So if the endpoint you want to compare in the study is relative reduction in seizures, it is much easier to demonstrate a higher relative reduction in people who are having a far higher number of seizures than in patients who are having a much lower number of seizures to begin with.
Think of a weight loss study that is measuring percentage of body mass that is lost with a specific therapy – and now imagine that in the treatment group, the average weight was 400 pounds, and in the placebo group the average weight was only 200 pounds. No matter what the intervention is, you can predict with high accuracy which group is going to have the higher percentage weight loss – the 400 pound group, of course, because a much lower percentage of their weight comes from lean body mass, and they have both more relative weight and more absolute weight to lose.
Put another way: It is much easier to measure a disease reduction in people who had more disease to begin with.
So now let's move back to the LBPH study, where the treatment group at baseline had about 40 seizures a day, and the placebo group at baseline only had about 20 a day. I think you could have predicted, once the trial participants were chosen, who was going to have the greater percentage reduction, without even knowing much about the intervention – the 40-seizure-a-day group, of course.
And that's what you saw – about a 50% reduction in the 40-seizure-a-day group (which they are calling the treatment group — a true but incomplete descriptor, as this designation ignores the study’s baseline imbalance), and about a 20% reduction in the 20-seizure-a-day group.
IS THERE AN ALTERNATIVE MANNER OF INTERPRETING THE RESULTS?
Yes.
In fact, I can think of two observations that I have not seen or heard in public discussion.
For one, you saw a 20% reduction in the placebo group — the patients who received no new therapy at all — which tells you how much better the patients in this trial did simply by getting enrolled and receiving care with close follow up.
The other thing is that, at the end of the trial, the placebo group was having less seizures a day than the treatment group: if you you’re having 40 seizures a day and they are reduced by 50%, you're down to 20; and if you’re having 20 seizures a day and they are reduced by a fifth, you are down to 16. In actual numbers from the trial, the treatment group got down to 18 seizures a day, and the placebo group was “only” having 16.
And so LBPH seems to want to have it both ways: They want to focus on the relative reduction in seizures in each group, instead of focusing on the total number of seizures in each group — while simultaneously not discussing the rather significant baseline imbalance in seizure activity in the two groups as the trial begun.
In the end you may have a greater percentage reduction in the treatment group – but you can see how much the baseline imbalance affected that (and perhaps even assured that).
SIZE MATTERS
And then there is the simple matter of the number of patients in each study group: there were nearly five times as many patients in the treatment group as in the placebo group (43 v. 9). I can't promise you what would have happened to the placebo group if it had a greater allocation/enrollment, but with 9 patients it didn't have all that much room to demonstrate trends.
And here's another notable baseline imbalance: the treatment group had more than 10% of its patients with a diagnosis of Dravet Syndrome, which is a rather severe epileptic/developmental syndrome (life expectancy generally under 10), and the placebo group had ZERO patients with Dravet Syndrome.
Which means that the disease with the greatest severity, and the greatest seizure frequency, was excluded from the placebo group – which gave the placebo group a lower chance at achieving a greater percentage reduction. So this was another large imbalance in the baseline – the patients in the placebo group did not even have the same diseases as those in the treatment group, let alone one of the diseases in question.
Finally, you have to look at the adverse event and dropout rate. The placebo group had none (although it was only nine patients large, but it still had zero dropouts). The treatment group had 9 dropouts due to adverse events – 20% of the group dropped the therapy, including 7 during the induction/titration phase. I don't think that is the death-knell of bexicaserin, but you do have to acknowledge that there was a 20% dropout/AE rate with the treatment, and 0% with the placebo.
But the bigger issue, to me, is the differences between the study group baselines. This whole trial is colored by baseline imbalance, and though on a charitable day I might be willing to acknowledge this study as a proof of concept, I would not go so far as to say that you've got a home run here. I don't think any intellectually honest scientist would say that they have proof of efficacy, based on how this trial was designed and conducted.
This type of design will not cut it in mid-stage and pivotal trials.
Quote of the Day
“The cleverest of all, in my opinion, is the man who calls himself a fool at least once a month.” — Fyodor Dostoevsky
Mark the Calendar
JDRF Ride to Cure Diabetes, October 24-27, Death Valley, California. JDRF (formerly the Juvenile Diabetes Research Foundation) hosts five 100-mile bicycle rides a year as fundraisers for Type 1 diabetes research and advocacy. So what would could be better than a Century Ride on the Hottest Place on Earth? The good news: It’s flat. And below sea level. You won’t have to deal with altitude.