A few minutes on Thursday morning could decide Biogen's future. At a hotel along the San Diego Bay, researchers will present more details from a pair of clinical trials that tested the company's most prized Alzheimer's drug in patients with early stages of the disease.
If convincing, the results would strengthen Biogen's case that it has the first-ever treatment for the underlying cause of Alzheimer's, a breakthrough potentially worth billions of dollars.
But that's a big if.
Biogen has drawn skepticism and at times criticism for how it analyzed the two trials, titled ENGAGE and EMERGE, and for its recent decision to seek approval based on their data. While some analysts and doctors seem to have already decided whether or not they support Biogen's rationale, Thursday's presentation — which kicks off the first full day of the 12th annual Clinical Trials on Alzheimer's Disease conference — offers a chance to provide much-needed answers.
A good barometer for how transparent and robust Biogen's answers are will be its market value, considering that value has shifted by tens of billions of dollars twice this year because of EMERGE and ENGAGE. Depending on the strength of the fuller dataset, there could be a third dramatic swing around the corner.
Industry watchers will be listening closely for clarity on several issues, including these three:
1. How Biogen accounted for missing data
ENGAGE and EMERGE had different outcomes. But one thing they had in common, besides their identical design, was that a large number of patients didn't make it through the 78-week study period. Of the 1,647 patients enrolled in ENGAGE and 1,638 patients enrolled in EMERGE, 34% and 40% hadn't completed their treatment by March 20 — when Biogen halted both studies because they looked unlikely to succeed.
The high non-completion rates make it more challenging to evaluate the trials, particularly EMERGE, which is the one Biogen claims as positive and the basis for approval. In that trial, Biogen said patients who completed the full course of treatment on the higher dose of its drug, called aducanumab, showed a 23% greater reduction on a dementia rating scale than patients given placebo.
Interestingly, the company also reported a 23% greater reduction for the entire group that received the higher aducanumab dose, regardless of whether they completed the trial — a group referred to as intent-to-treat, or ITT.
Evercore ISI analyst Umer Raffat sees the 23% figure reported for the broader group as the result of Biogen inferring patients who didn't complete their aducanumab treatments experienced a similar level of benefit as those who did.
Indeed, Biogen and its co-development partner Eisai have been using a statistical model that works on those kinds of inferences. Yet the model typically operates on the assumption that data from non-completer patients are missing at random. Raffat views this modeling choice as a potential problem for Biogen, since the data weren't missing at random, but rather because ENGAGE and EMERGE were stopped early.
Brian Abrahams of RBC Capital Markets expects a main point of interest at CTAD will be the statistical methods Biogen used, and whether those methods were always part of its analysis plan. If not, they might provide more ammunition to critics who say Biogen just molded the data until it looked positive.
Follow-up data on patients who didn't complete EMERGE will also be important, according to Abrahams, because the closer they were to week 78, the safer it is to make assumptions about their responses to aducanumab.
2. The reliability of the EMERGE results
Clinical trials succeed or fail based on something called p-value, which expresses the probability that any observed benefit is due to the drug being tested rather than random chance. The smaller the p-value, the more confident researchers are in the trial's results. Typically, p-values of less than 0.05 indicate a statistically significant finding.
In the case of EMERGE, Biogen said the trial succeeded because the higher aducanumab dose met statistical significance. The p-value reported for the higher-dose ITT group was 0.010, while the p-value for just the patients who completed the study was 0.031. In other words, according to Biogen, it's fair to conclude the 23% reduction compared to placebo was because of aducanumab.
It's not clear what threshold for statistical success Biogen initially set for ENGAGE and EMERGE. But, as Raffat and others have pointed out, the company had to make assumptions when it was putting together the data analysis that ultimately deemed EMERGE positive. A looming question, therefore, is whether the trial would still have hit statistical significance if those assumptions were changed, even slightly.
Some don't seem to think it would.
"Our view is that there isn’t a single statistician within the FDA that will support the reliability of these results, and to highlight that, we believe the agency will likely be able to easily find analyses that support all sorts of findings," Brian Skorney, an analyst at Baird, wrote in a Dec. 2 report.
"The fact is, the prospectively defined pivotal program failed, making any further analyses exploratory in nature and any p-values reported nominal," Skorney added. "To make claims otherwise is statistical malpractice."
3. How well the data reflect a patient benefit
If everything Biogen reported so far holds true — EMERGE remains positive, statistically significant, and supported by 100 or so ENGAGE patients who also seemed to respond to aducanumab — the results still need context.
A crucial consideration is whether aducanumab proves clinically meaningful, or basically that the drug has real-life, noticeable benefits for patients.
Biogen argues that reducing clinical decline by 20% or more compared to placebo represents a clinical meaningful response. But whether a 20% reduction, or 23% reduction for that matter, actually improves patients' disease remains up for debate.
Skorney doesn't foresee it being substantial. The Baird team estimates a 23% reduction equates to a roughly 0.4-point change on that dementia rating scale used in EMERGE and ENGAGE. A new published article in The Lancet also estimates the change was equivalent to about 0.4 points.
That difference is "of uncertain clinical relevance, especially in light of the lack of effect in the ENGAGE trial," wrote Lon Schneider, a professor of psychiatry and neurology at the University of Southern California, in that Lancet article.
Comparatively, Eli Lilly's solanezumab, another failed Alzheimer's drug that works similarly to aducanumab, showed a 0.34-point difference versus placebo. Years before that, a now generic drug called donepezil was found to have a 0.5-point benefit in a 24-week clinical trial. It's worth nothing, though, that neither of those studies used the dementia rating scale as their primary endpoints.
"This is a drug that is widely viewed to be of minimal benefit," Skorney wrote, referring to donepezil, "yet, based on the EMERGE results, a patient would need to be on aducanumab for almost [four times] the amount of time to get the same benefit."
Unlike donepezil, aducanumab is meant to be disease-modifying by clearing amyloid plaques from the brain — something Biogen's results showed aducanumab capable of doing. Whether eliminating the characteristic plaques is the right approach to changing the trajectory of Alzheimer's, though, is a question that's evaded researchers for decades.
A deeper explanation of how aducanumab offers clinically meaningful benefits may help Biogen ease another concern: safety. As with many other amyloid-targeting Alzheimer's drugs, a significant portion of patients who took aducanumab experienced a side effect known as ARIA, or amyloid-related imaging abnormalities.
One type of ARIA causes brain edema, or swelling. While such an effect isn't to be taken lightly, Howard Fillit of the Alzheimer's Drug Discovery Foundation notes that these edema cases typically resolve themselves.
Fillit, who serves as the ADDF's chief science officer, already views aducanumab as safe. During the CTAD presentation, he said he'll be more focused on longer-term efficacy data, how patients responded depending on their dose, and any sort of additional outcomes data, in particular functional outcomes.
"It's pretty clear that aducanumab removes amyloid plaques in the brain," Fillet said in an interview with BioPharma Dive.
But "it's not just whether they got a p-value, but what the effect size is," he added. "I've seen some interpretations of the data where the effect size isn't really that great. And if that's true, then there'll be a lot of questions about the ultimate clinical value of the drug."