Eli Lilly: Experimental Alzheimer’s drug shows some benefit (AP)
Eli Lilly’s Solanezumab Is the Alzheimer’s Drug Equivalent of the Internet Dress (TheStreet)
Lilly says drug slows Alzheimer’s in patients with mild disease (Reuters)
Lilly’s Alzheimer’s Drug May Slow Patients’ Decline (The Wall Street Journal)
Biogen, Lilly Alzheimer’s Data Spark Street Debate (Investor’s Business Daily)
Biogen: Come On, the Results Weren’t That Bad (Barrons.com)
Biogen Alzheimer’s drug data falls flat, Lilly gets slight bump (Reuters)
Alzheimer’s Trials Offer Few Answers for Investors or Patients (Bloomberg)
Lilly Alzheimer’s Drug Shows Benefit in Those Who Start Earlier (Bloomberg)
Promising Alzheimer’s Drugs Disappoint With Incremental Data (Forbes)
The responses to the data releases for solanezumab and aducanumab were not a model of consensus. As is not unusual, the ramp-up of expectations prior to the solanezumab and aducanumab supplemental data had been unrealistically steep, and the aftermath featured a broad spectrum of divergent responses and interpretations, though the predominant themes were vague disappointment and confusion. When one media outlet has to describe results in terms of “weren’t that bad”, one is not in the presence of definitive success. The expectations were overheated, given the limited nature of what to be unveiled: Biogen’s addition of 52 week data for a dose-level cohort that had enrolled just 30 patients; Lilly’s report of results from an extension of two failed studies for solanezumab, selecting subgroups from those Phase III populations, then comparing outcomes between patients who had been on solanezumab with those who were on placebo until the extension component began, eighteen months later. The number of patients who completed Lilly’s extension study was 626, which might have been impressive, other than for the fact that the trial began with 1314 patients, and the extension component began with 1018. With the attrition rate more than 50% from the beginning of the trial, one must wonder whether the 48% who stayed the course were a representative sample of the subgroups Lilly had selected posthoc, who may or may not have been representative of mild Alzheimer’s (eighteen months later).
Consider the magnitude of change in the Lilly trial. The extension of course did not have a control group, the trial was predicated on the premise that if the drug is disease-modifying, those patients who started on the drug later would remain cognitively ‘behind’ those who started earlier. This is an experimental protocol that has yet to be embraced by the FDA (though Paul Leber played a role in its devise), and it may never be: It can be argued that the two groups are inherently different by virtue of one beginning treatment 18 months later than the other, perhaps having reached a different stage (and slope) of their disease course. Not everyone accepts this divergence as evidence that the disease-course has been modified. But just to take it at face value for the moment, the graph of the results showed the two groups mirroring each other, just slightly separated, for two years on the ADAS-cog. The divergence was statistically significant at all timepoints except the last one, where the p value shot up to .169. That could reflect the shrinkage of the population and/or greater variance in the test results for those who remained.
Even if Lilly’s framing of the results has merit, the mean difference in ADAS-cog score decline one year into the extension study is just 1.91 points. The FDA’s general rule of thumb is that a 4 point difference on the ADAS-cog is clinically meaningful. The ADNI group reported that mild AD patients receiving cholinesterase inhibitors declined an average of 9.25 points over two years; another 18 month study found a mean decline (in the placebo group) of 6.44 points. These studies point to a 4-4.6 point annual decline rate for this population, which means (with the caveat that decline rates in Alzheimer’s are not linear, and may vary between genders, with women declining more rapidly) that a 1.91 pt difference is a bit less than one would expect a modal patient to decline in six months. Similarly, on the MMSE, Lilly’s slide (we await the imminent publication of the trial report) showed what looked to be less than a one-point difference between the two groups at two years into the extension trial, 3.5 years into the overall study: Three points is considered to be a significant score change on the MMSE, or roughly, just under the decline that would be seen over an eighteen month period. Which would again point to an treatment effect tantamount to ‘six-months worth’ of progression. It is possible that the impact might grow over time, but then again, it is possible that some of this is ‘noise’, perhaps partly due to the notable attrition rate in the study.
Lilly may yet achieve statistical significance in the ongoing ‘Expedition 3’ trial, which uses a traditional placebo-controlled protocol, but even if they do, there is nothing in these results that suggests that clearcut clinical significance is to be assumed. But it is possible that even a minimal clinical effect might be accepted as better than nothing, provided that no safety concerns emerge (the original Phase III data showed just a 1% likelihood of vasogenic edema).
So here is a headline that captures the gist of the findings from the solanezumab data:
Solanezumab May Do Something, But Not Much–But It Might be Better Than Nothing
We had initially thought that both mAbs had an equally lackluster day today, but that is not the case. First, it looks like 26 patients in total completed all cognitive test sequences in the long-awaited 6mg/12 month group. That is not a cohort size about which much is generally said when it comes to Alzheimer’s. Anomalies in the readout could reflect some outlier patient(s) whose impact would be greatly magnified by the small n. But when Biogen presented the results, they claimed that they fell in line with the preliminary data from last March, which is not completely true. The rate of decline on the CDR-SB did indeed place the 6mg group between the 3mg and 10mg groups, but it was much closer to the 3mg group. The neat dose-response relationship that had initially surfaced in March vaporized: The MMSE score decline for the 6mg group was almost identical to the very limited effect seen from 1mg, while the 3mg and 10mg groups clustered together in an oasis of statistical significance. The impact on amyloid plaque was linear, but since that did not consistently line up with functional effect, this raises more questions than it answers. Finally, the incidence of vasogenic edema in the 6mg group (37%) was nearly the same as in the 10mg group (41%), far above that seen with 3mg (6%). So the hope that the 6mg dose might ‘walk the tightrope’ and somehow provide 10mg dosing efficacy along with 3mg dose adverse effects was not fulfilled, at least not in this trial. While Biogen tried to reassure observers that the vasogenic edema findings were not particularly problematic (“typically resolved within 4-12 weeks”), the fact is that while 78% were “mild to moderate”, this means 22% were more severe. In a trial that enrolls 166 patients and makes brain scans easily available, that is manageable. In a general Alzheimer’s population, perhaps not. For example, if 100,000 patients were to receive aducanumab at the 6mg dose, and if the incidence rate in this trial is predictive, 37,000 could develop vasogenic edema, with 7500 of those cases being severe. That becomes a clinical monitoring and management challenge that could outweigh what looks like a modest-at-best treatment effect: Even the most seemingly efficacious dose (10mg) achieved just a 2.25 point difference on the MMSE decline rate at 12 months, on a test where three points is the minimum considered to be clinically meaningful. On the CDR-SB, where mild Alzheimer’s patients decline 1.4 points annually, the 10mg dose approached that level with a 1.24 point effect, the 6mg dose had just a .76 point impact, close to the ‘six-month’s worth’ noted for the MMSE. Again, these are tiny dose cohorts, and may not mean anything, but today’s additional data only beclouded the issue for aducanumab.
Thus, because of the added concern about vasogenic edema, the aducanumab headline reads like this:
Aducanumab May Do Something, But Not Much–It Might be Better Than Nothing, But Possibly Not