Pages

Wednesday, April 28, 2010

MermaiHD Madness: NeuroSearch's Statistically Significant Snafu


Call it the Curse of the Little Mermaid.

In February, NeuroSearch trumpeted statistically significant Phase III results from its MermaiHD trial lead project Huntexil in Huntington's disease. The positive data sent the company's shares soaring. The Danish biotech's value nearly tripled, and analysts suggested that the data supported launch of the drug in 2011. The company spent the past few months topping up the good news: additional data presented at conferences, new information supporting disease modifying properties of the drug, etc.

Neurosearch was riding high -- a company that has had difficulty over the past decade securing positive Phase III results after promising proof of concept data finally got over the hump.

And then, last week, Denmark's famous Little Mermaid statue was removed from Copenhagen Harbor and sent to Shangai, where it will take pride of place at the Danish Pavillion during the 2010 World Expo.

Mermaid gone (a replica in Tivoli Gardens just ain't the same, we guess), and MermaiHD is all of the sudden showing cracks.

This morning NeuroSearch put out a release to clarify the MermaiHD results. That missive shaved 40% off the company's value almost immediately, though the share price has slowly recovered a bit as the day wears on and executives have had the chance to explain themselves. The upshot? It turns out that MermaiHD's positive results around the primary endpoint of modified Motor Score (mMS) weren't statistically significant, after all.

That's right. "Additional data assessment" turns out to skew the Phase III results just slightly, but enough to upend the statistical significance of MermaiHD's primary endpoint. "Further assessment of data from the study shows that the significance value for the primary study endpoint, the modified Motor Score (mMS) of p= 0.042 did not meet the pre-specified level of p [less than] 0.025," says the company's release.

Wait, can you say that in a much longer, more detailed way?

The conclusion regarding the primary endpoint, the mMS with a significance level of p [less than] 0.02, which was communicated as part of the top-line results, was based on a clinically relevant baseline covariate adjustment for differences in patients' genetic disposition, i.e. the length of CAG repeats (CAGn) in the diseased gene sequence. This adjustment is judged to be clinically important and appropriate in ensuring a more meaningful representation of the data set. Based on this assessment, the primary endpoint for the MermaiHD study was concluded to be met (p [less than] 0.025). The adjustment for individual differences in patients' CAGn x treatment was pre-specified in the study protocol as a sensitivity analysis but not as part of the main effects model for the primary analysis. In view of this, the statistical results have been re-assessed, demonstrating a formal p-value of 0.042 for the primary endpoint, the mMS, and consequently indicating that the study did not rearch the p [less than] 0.025 significance level (Bonferroni adjustment) as pre-defined in the study protocol.
Get it, statheads? But what does it all mean? NeuroSearch maintains that its regulatory strategy won't change and that the new data "doesn't change the overall picture" for the drug. Analysts who were able to wrap their heads around the implications for Huntexil during the company's results call this morning seem to agree that this isn't the end of the line for the Huntington's drug -- thought it may complicate the regulatory process a bit. Results from a second Huntexil Phase III, the so-called HART study, are expected in Q3.

Let's hope for NeuroSearch's sake, the Little Mermaid is back from China by then.

image from flickr user celesteh used under a creative commons license

No comments: