One failed trial followed by a positive one equals an 11% increase in stock price.

Technically Neurocrine Biosciences (NBIX 0.23%) was up 90% today, but it dropped so much after the first trial, the biotech is only sitting on a relatively modest gain from where it was before results from the first trial were announced.

The two clinical trials tested Neurocrine's NBI-98854 in patients with tardive dyskinesia, which causes involuntary movements like lip smacking or blinking. In the first trial, patients saw a decrease of 3.5 points on a modified Abnormal Involuntary Movement Scale, or AIMS, after six weeks of treatment compared to a drop of 2.2 points for the placebo group. A lower AIMS corresponds to an improvement, and in both cases, the scores were determined by an independent reviewer.

Getting rid of the placebo effect
While a 3.5-point reduction in AIMS is pretty substantial, it wasn't statistically significant because of the rather large improvement in the placebo group. The placebo effect is one of the main reasons clinical trials that use subjective scales fail. Lexicon Pharmaceuticals' (LXRX 0.65%) LX2931, for instance, produced a solid 60% response rate in rheumatoid arthritis patients, but 49% of the placebo group also had a positive response, so there's no way to know how much of the improvement is due to the drug.

While Lexicon put LX2931 on the back burner, Neurocrine pressed on with NBI-98854 because some of the patients in the earlier trial were given a 100 mg dose for the first two weeks, which produced a statistically significant improvement in AIMS after just two weeks. Patients were then switched to a 50 mg dose like the rest of the patients in the treatment group, so there weren't six-week data for that cohort.

For the second trial, Neurocrine made several changes based on what it had learned. A different modified AIMS was used, which was scored by movement disorder neurologists. Importantly, to reduce reviewer bias, the reviewers didn't know whether they were watching a video of a test to establish a baseline AIMS score or one after the six-week dosing period was complete.

Those differences seem to have cut out nearly all the placebo effect, which decreased by just 0.2 points on the new AIMS scale. Patients who got NBI-98854 -- starting at a 25 mg dose, with most escalating to 50 mg and then to 75 mg for the final two weeks -- saw their AIMS decrease by 2.6 points, which was highly statistically significant when compared with placebo. When Neurocrine used another measurement of symptoms, called the Clinical Global Impression-Tardive Dyskinesia, there was also a statistically significant improvement compared to placebo.

Moving forward
Neurocrine Biosciences plans to set up a meeting with the Food and Drug Administration to work out details of a longer phase 3 trial -- perhaps two -- which will hopefully start in late summer.

To prove that the results from the second trial weren't a fluke, Neurocrine plans to go back and rescore the video tests from the first clinical trial using a randomized order and the second more-sensitive scale. If NBI-98854 is really working, the adjudicated data from the first trial should look better than it first appeared, although perhaps not as good as the second trial which escalated patients up to 75 mg.

I take from the modest gain since September's highs that investors aren't fully convinced that NBI-98854 really works on tardive dyskinesia, which seems reasonable with the company only batting .500. Shares could go higher if the adjudicated data from the first trial looks promising, but results from a large phase 3 trial -- more than a year away -- will be the definitive move.