Chinese Data Torture

Drip…drip…drip. The solanezumab results continue to trickle out. In August, Lilly announced that the two Phase III trials had missed their endpoints, but data from mild Alzheimer’s patients pointed to a possible treatment effect on cognition, but not on functioning. The release was very short on details, like effect sizes and p-values, as Neurogram has previously discussed. Now, in October, results from an independent (ADCS) data analysis were presented at the American Neurological Association meeting. The evidence for an effect on the rate of cognitive decline in mild AD was shown in more detail, with the pooled data from both trials showing a 34% reduction in the decline as measured by the ADAS-cog. However, the actual difference in ADAS-cog scores between the two groups was just 1.4 points–more on that later. Lilly’s initial report, and a more detailed press release that they issued during the ANA meeting (though they did not present their own analysis there) had stated that the solanezumab group did not show any difference on a functional measures (ADSC-L)–but the independent analysis noted that there was a trend towards benefit, with a p value of .057. Perhaps Lilly’s attorneys forbade the use of the word ‘trend,’ but this is another example of statistical ‘significance’, which is arbitrarily binary, obscuring rather than revealing.

Unfortunately, Lilly’s analysis and the independent analysis are not available for side-by-side comparison. The evidence thus far points to some kind of treatment effect, only in mild AD patients, but at a level (1.4 ADAS-cog points) which is so tiny that statistical significance does not necessarily say anything about clinical significance. Even the cholinesterase inhibitors beat solanezumab on that score, and EnVivo‘s EVP-6124 outstripped it markedly, albeit over just six months. A symptomatic treatment that only achieved a 1.4 pt difference would be terminated immediately.

But the more important point is–does this validate the beta-amyloid hypothesis, and show that if we only intervene early enough (provided we can identify the vulnerable population) that a clinically significant impact on progression will be shown? The consensus in the AD community trended towards the affirmative, but others are more cautious. Concluding that there is an impact upon the underlying biology of the disease via beta-amyloid could require confirmation via the biomarker data obtained from the two trials. If it does not line up, one would have to conclude that the cognitive/functional data either are fluky, or reflect some alternative mechanism of effect. So, what does that biomarker data tell us?

Nothing yet. The presentation in Boston reported only that solanezumab had an effect on some biomarkers, not others. Biomarker data awaits yet another meeting, in Monte Carlo, in November. Other than allowing a certain amount of irony given the setting for disclosure (the odds are far better at the baccarat table), there really is no excuse for this. One can justify a topline data release in August, even as further analyses are done. But clearly, the biomarker results are known, and they should have been released at the same time as these somewhat (albeit incomplete) more detailed cognitive data. We understand the appeal of traveling to high-end locations like Monaco, but this is not supposed to be a HBO series that maximizes viewership by leaving the resolution ambiguous, guaranteeing that the audience will return the next season. They have the full attention of the neuroscience community– the trickle-down approach to data release is unjustified.

This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s