‘Reports of My Death Are Greatly Exaggerated.’ Signed, Aducanumab
Quick Links
On October 22, Biogen stunned the Alzheimer’s field by announcing that aducanumab—presumed dead last March after failing a futility analysis—appears to have worked in one of its two Phase 3 trials, after all. Based on the results of a new analysis, and interactions with the Food and Drug Administration, Biogen will file for regulatory approval in early 2020.
- Aducanumab rises from the dead based on analysis of more data from Phase 3.
- In the larger dataset, the EMERGE trial was positive, ENGAGE was not.
- Citing feedback from the FDA, Biogen believes the data support a regulatory filing.
Why the revival? In a quarterly investor phone call, Biogen CEO Michel Vounatsos said the interim futility analysis was flawed and did not adequately take into account the effect of two late protocol amendments that boosted the number of people receiving the highest dose of this biologic drug. A new analysis conducted since March 2019 included three more months of data, as well as data from the participants who did not complete the full course of treatment. It showed that one of the two trials, called EMERGE, in fact met its primary and secondary endpoints. Oddly, the identical ENGAGE trial, which started one month earlier, was a tad larger, yet had slightly fewer people who took an uninterrupted course of the maximum dose, remained negative. However, an exploratory subgroup analysis of ENGAGE participants who received 10 or more consecutive high doses did show a slowing of cognitive decline similar to that seen in EMERGE, Biogen researchers said.
“We were surprised when we learned about the potential implications of these additional data,” Vounatsos acknowledged. “The path to developing breakthrough treatments is not always direct and straightforward.”
Alzheimer’s researchers largely hailed the news. “These results breathe new life into this therapeutic strategy,” Colin Masters at the University of Melbourne, Australia, wrote to Alzforum. Paul Aisen at the University of Southern California, San Diego, agreed. “Overall, this is enormously encouraging for the field. While we need to fully digest these complex data, the results seem to support the amyloid hypothesis—very good news for multiple drug development programs underway—and may even indicate that disease-modifying therapy for AD is nearing the clinic,” Aisen wrote. “This is an injection of optimism for the research field and patients,” wrote Kaj Blennow of the University of Gothenburg. “I missed the original resurrection, so I am glad to be around for this one,” quipped Stephen Salloway of Brown University, Providence, Rhode Island, adding, “This is a great development.”
At the same time, the researchers struck a note of caution, saying they would reserve judgment until they see a fuller dataset. “There is not enough information here to assess the validity of the analysis,” Lon Schneider of USC, Los Angeles, told Alzforum. Nearly 20 researchers contacted by Alzforum echoed Schneider’s sentiment, though many spoke only off the record. Additional data will be presented at the 12th Clinical Trials on Alzheimer’s Disease conference, scheduled for December 4–7 in San Diego.
Aducanumab initially excited the field when it appeared to remove brain amyloid and slow cognitive decline in the Phase 1b PRIME trial (Mar 2015 conference news). Some researchers raised concern about the underpowered cognitive findings (Sep 2016 news). The data appeared in the journal Nature (Sevigny et al., 2016), and Biogen launched a Phase 3 program. In March 2019, Biogen stopped both those trials after a futility analysis predicted they would miss their primary endpoints (Mar 2019 news).
This futility analysis was based only on data from the 803 participants in EMERGE and 945 participants in ENGAGE who had completed the full 18-month course of treatment as of December 2018, representing about half the enrollment of each trial. By March 2019, when the trials were stopped, an additional 179 EMERGE participants and 139 ENGAGE participants had completed treatment. Biogen’s subsequent analysis of the larger dataset included these additional completers, as well as data from the rest of the participants who had not finished the intended course of treatment. EMERGE enrolled 1,638 people; ENGAGE, 1,647. A third of the participants received placebo, the remainder drug.
In each trial, about half the participants on aducanumab were randomly assigned to titrate up to a low dose of drug—3 mg/kg for ApoE4 carriers, 6 mg/kg for noncarriers. This difference was because ApoE4 carriers are more susceptible to ARIA, the fluid retention in the brain that accompanies treatment with many amyloid-removing therapies. In the high-dose group, ApoE4 carriers initially titrated up to 6 mg/kg; noncarriers to 10 mg/kg. However, in March 2017, about 18 months into the trial, the protocol was amended to allow ApoE4 carriers to titrate up to 10 mg/kg, as well. This was based on accumulating data from several studies suggesting that ARIA is a manageable side effect that usually resolves without harm.
For the same reason, the researchers had already made a prior protocol adjustment. Initially, participants who developed ARIA would stop dosing for a few weeks until it resolved, then resume at a lower dose. In July 2016, about a year into the study, participants with ARIA were allowed to resume dosing at their original level. Both of these protocol adjustments increased the number of participants receiving the highest dose. However, the futility analysis proceeded as originally conceived, and was not adjusted to incorporate the effect of these amendments on dosing.
The amendments affected more EMERGE than ENGAGE participants, because the latter trial started enrolling a month earlier and remained ahead throughout the trials, said Samantha Budd Haeberlein, who leads late-stage clinical development at Biogen. Overall, more than 10 percent fewer ENGAGE than EMERGE participants received 10 consecutive high doses, which Biogen set as the benchmark for sufficient exposure to drug. Budd Haeberlein did not provide precise numbers for how many participants in each trial reached this benchmark, promising to show those data at CTAD. Nor did she explain by what rationale Biogen selected this benchmark.
Analysis of the more recent, larger dataset suggested that duration of treatment at the high dose was the key factor. In addition, Budd Haeberlein said, interruption of treatment played a role. In EMERGE, participants on the low dose had a trend of declining more slowly than those on placebo on the primary outcome, the CDR Sum of Boxes, but this was well shy of statistical significance. Participants on the high dose declined 23 percent more slowly than those on placebo, with a significant p value of 0.01. Secondary endpoints were similar. The high-dose group declined about a quarter less on the ADAS-Cog13, a cognitive battery, and up to 46 percent more slowly on the ADCS Activities of Daily Living, a caregiver assessment. Both results were statistically significant for both the intention-to-treat group, i.e., all participants who received some doses of aducanumab, and for those who finished the 18-month course. Decline on the MMSE reached significance in the high-dose completers, but not the whole cohort.
In ENGAGE, on the other hand, decline on the CDR-SB slowed a tad only in the low-dose group, missing statistical significance. The three secondary outcomes were similar, with some slowing of decline on each that missed significance. The high-dose group posted a slight trend toward worsening on the primary endpoint and the MMSE, and a slight benefit on the ADAS-Cog13 and the ADCS-ADL, which fell short of significance. Budd Haeberlein noted that these two measures are more sensitive indicators of change than the CDR and MMSE. Analysts on the investor call questioned why the low-dose group appeared to reap more benefit than the high-dose group in this study. Budd Haeberlein replied that the high-dose group had more interruptions due to ARIA and also more dose changes due to the protocol amendments, making dosing less consistent for these participants.
An exploratory analysis of just ENGAGE participants who received 10 or more doses of 10 mg/kg without interruption—97 people on aducanumab compared with the 336 on placebo by 78 weeks—hinted at a slowing of decline on the CDR-SB similar to that seen in EMERGE.
“This shows us that we have to get to a high dose to have an effect. Intermediate dosing, at least in an 18-month trial, is not enough,” Al Sandrock, who heads research and development at Biogen, said in the investor call. Sandrock noted that the Phase 1b PRIME study had shown some slowing of decline at 3 mg/kg and 10 mg/kg, but not at 6 mg/kg. At the time, the 6 mg/kg dose was believed to be an outlier. Now, Biogen researchers believe that the 3 mg/kg dose was the outlier in that study. “We believe 10 mg/kg is the correct dose for efficacy,” Budd Haeberlein said.
Biomarker findings, as presented, seem to support this. On amyloid PET, the researchers reported a decline of about 0.16 SUVR in those on low doses of aducanumab, 0.24 in the high-dose ENGAGE group, and 0.27 in the high-dose EMERGE group. The placebo group held steady. “We do believe the difference in amyloid reduction between groups is significant,” Budd Haeberlein said. “PET measurement of amyloid reduction is a very sensitive indicator of dose.” Notably, the ENGAGE subgroup with high exposure to aducanumab cleared as much amyloid as the EMERGE high-dose group.
Researchers wanted more data on this point. “It would be helpful to know how effectively lower doses led to changes in amyloid PET over time compared with the 10 mg/kg subset,” Douglas Galasko at the University of California, San Diego, wrote to Alzforum. He was a site investigator for the aducanumab Phase 3 trials. Others expressed some reservation about white-matter binding in the amyloid PET scans.
In the call, Biogen showed some data on CSF tau in subgroups of 15 to 29 people. P-tau nudged downward in both EMERGE and ENGAGE, in both low- and high-dose groups, by up to 19 picograms/ml and with statistical significance. Total tau dropped by up to 160 picograms/ml in EMERGE, though curiously not on the high dose in ENGAGE; these changes missed statistical significance. “The CSF p-tau data are quite encouraging, supporting the idea that reducing amyloid has a beneficial effect on downstream neurodegeneration,” Aisen noted. Other biomarker researchers called the numbers hard to evaluate without baseline levels given. Several commentators told Alzforum they would like to see spaghetti plots of how individuals fared on CSF tau, amyloid PET, and CDR-SB, saying that this way of showing the data would help make sense of this heterogeneous dataset.
Several researchers Alzforum spoke with were troubled by the difference in the EMERGE and ENGAGE outcomes, questioning if the slightly higher number of people on the highest dose in EMERGE was enough to account for it. Several said they wanted to see an analysis of ApoE4 carriers versus noncarriers, and the rates of ARIA in each group. ApoE4 carriers would be expected to have more ARIA and more dosing interruptions. Howard Feldman at the University of California, San Diego, noted the importance of knowing the standardized effect size, so that aducanumab can be compared with other interventions.
Overall, most researchers cautiously agreed that what they can see so far of the new analysis appears sound, and that the futility analysis was likely wrong. “This is not the first time I’ve seen a company get burned by futility analysis,” said Marwan Sabbagh, who leads the Cleveland Clinic Lou Ruvo Center for Brain Health in Las Vegas. Roche’s gantenerumab failed futility analysis in the Phase 3 SCarlet RoAD trial (Dec 2014 news). That nearly derailed the program, before a deeper data analysis indicated that differences between fast and slow progressors in the trial might have masked a possible treatment benefit among a subgroup of participants with the highest exposure (Nov 2015 news). “We learned from our early studies with low doses of gantenerumab that exposure is what drives amyloid lowering as well as clinical signals,” Rachelle Doody of Roche wrote to Alzforum (full comment below).
“I do think such analyses can be risky, and result in a loss of statistical power. This is particularly true when treatment protocols are amended midstream, so that early data may not be representative of later data,” Aisen wrote. Salloway noted, “With a futility analysis, you can end up killing a promising drug by mistake.”
Some commentators recommended that futility analyses be applied with more nuance. Galasko noted that the timing of a cognitive and clinical response to amyloid removal is still so poorly understood that it might be better to base futility analyses on biomarker change, not on cognitive/clinical outcome measures. Along a similar vein, Randy Bateman at Washington University in St. Louis wrote to Alzforum, “After taking years for a program to get to Phase 3 trials, I prefer to see more adaptive actions based on a futility, rather than complete shutdown of a program.” Off the record, others noted soul-searching at Biogen on the futility methods.
Sandrock touched on the issue himself, saying that in AD trials, the clinical benefit may lag behind amyloid removal. In PRIME, amyloid removal became significant at six months but a clinical signal did not appear until 12 months, he noted. Sabbagh said this makes sense, given that the removal of amyloid may gradually halt inflammation or tangle production, and this secondary effect may be what benefits cognition.
Many researchers lamented that the stoppage of these trials rendered the available aducanumab dataset messy and incomplete, with a mixed result of one positive and one negative trial. “That’s the biggest problem with futility analyses—unless the signal to stop is very clear, you never know what might have happened had you done the trial as originally planned,” Andrew Satlin at Intra-Cellular Therapies in New York City told Alzforum. Others called it upsetting that the sites put in tremendous work to launch these trials, recruit the participants, and then did not get to the end and a full dataset.
Biogen plans to call back participants in EMERGE, ENGAGE, and PRIME to resume dosing. Will they return? “Absolutely. People are clamoring to go back on aducanumab,” said Salloway. Experienced trialists develop a gut sense of whether a trial drug is “doing something.” While they do not talk up such anecdotal observations, some site clinicians privately expressed shock at the end of EMERGE and ENGAGE because in their minds, aducanumab was in this category. They were dismayed at having to stop, and are now pleased to be able to invite the participants back.
Biogen believes the current data support a regulatory filing. Sandrock noted that the full dataset was shown to the FDA, and the agency indicated that submitting an application was reasonable based on the data. He said Biogen is just beginning discussions with regulators in Europe and Japan.
Noting that the data from EMERGE show target engagement and slowing of disease progression, Sabbagh believes it might just be enough for approval. “I think the FDA has been looking for any shot on goal,” he said. If aducanumab were to be approved, it could transform the practice of Alzheimer’s medicine within a year or two, Sabbagh predicted. He believes the existence of an approved drug to remove amyloid would lead Medicare to approve coverage for amyloid PET scans. “This would be a game changer,” Sabbagh said.
With or without regulatory approval next year, this week’s news will intensify the field’s focus on developing blood and CSF biomarkers for screening, diagnosis, and monitoring treatment effects, said Blennow, who co-leads the Neurotoolkit and other initiatives to validate such markers (August 2019 conference news). Cheaper markers than PET are sorely needed.
Antibody therapies will be expensive, and regulatory approval may be a smaller hurdle to clear than insurance coverage in the U.S. and especially in countries with national health care systems. “Biologics, and also anti-sense oligonucleotides, will not be available to the masses. Prevention with small molecules is key,” Joachim Herz at the University of Texas Southwestern Medical Center in Dallas wrote to Alzforum.
Biogen scientists were unavailable to answer follow-up questions before this story was published. The slides of this presentation are available on the company website.
Biogen’s quarterly presentation led to a spike in its stock price by day’s end, while commentators talked over each other pumping up the story in the usual build-’em-up-tear-’em-down play cycle of TV analysts.—Madolyn Bowman Rogers
References
Therapeutics Citations
News Citations
- Biogen Antibody Buoyed by Phase 1 Data and Hungry Investors
- Paper Alert: Aducanumab Phase 1b Study Published
- Biogen/Eisai Halt Phase 3 Aducanumab Trials
- End of the RoAD for Gantenerumab? Roche Declares Prodromal Alzheimer’s Trial Futile
- Gantenerumab, Aducanumab: Bobbing Up and Down While Navigating Currents of Trial Design
- Proteomics Uncovers Potential Markers, Subtypes of Alzheimer’s
Paper Citations
- Sevigny J, Chiao P, Bussière T, Weinreb PH, Williams L, Maier M, Dunstan R, Salloway S, Chen T, Ling Y, O'Gorman J, Qian F, Arastu M, Li M, Chollate S, Brennan MS, Quintero-Monzon O, Scannevin RH, Arnold HM, Engber T, Rhodes K, Ferrero J, Hang Y, Mikulskis A, Grimm J, Hock C, Nitsch RM, Sandrock A. The antibody aducanumab reduces Aβ plaques in Alzheimer's disease. Nature. 2016 Aug 31;537(7618):50-6. PubMed.
External Citations
Further Reading
Annotate
To make an annotation you must Login or Register.
Comments
USC Alzheimer’s Therapeutic Research Institute
I am a member of the Aducanumab Steering Committee, so I continue to play an advisory role to Biogen.
Q: Based on the data you saw, does this new analysis look appropriate?
A: Yes.
Q: When ADCS did an independent analysis of Lilly's solanezumab data, that raised credibility and trust in how Lilly operated. Would you advise Biogen do that?
A: Biogen will be reviewing its data with its scientific advisory board, of which I am a member. It will also be sharing details at CTAD in December. I think these steps will go a long way toward credibility. While I believe the collaboration between the ADCS and Lilly was very valuable, I would not say such arrangements are required to gain trust and credibility.
Q: Was this futility analysis a mistake?
A: Companies invest enormous amounts in Alzheimer’s disease-modifying drug development, and they are usually eager to find ways to cut their losses through futility analyses. But I do think such analyses can be risky, and result in a loss of statistical power. This is particularly true when treatment protocols are amended midstream, so that early data may not be representative of later data.
Q: It sounds like Biogen underestimated how much effect the protocol amendments would have on the data, and did not consider the amendments in the futility analyses. Correct?
A: I think this is clear in hindsight.
Q: If this was your program, would you apply for FDA approval now or do a confirmatory trial?
A: Interpreting data from interrupted trials is challenging. Biogen has been meeting with the FDA to stay in sync. I have not yet seen all of the data, so I think it would be premature for me to weigh in on this question.
Q: Were the right outcome measures used for a trial in this population? Or do we now know enough to say the composites are better suited?
A: From what we see in the public data, there was general consistency across outcome measures. I do not see a reason to question the choice of measures.
Q: These protocol amendments, do they reflect the field’s initial intense worry about ARIA and then a subsequent sense that it was more manageable than initially feared? Here, dose was increased midway through these trials for ApoE4s. That’s the opposite of how BAN2401 Phase 2 was affected by the same issue, where midway through that trial, EMA requested E4s be taken off high dose, which unbalanced the groups.
A: In general, yes. It is quite challenging to balance risk against potential benefit when using amyloid-reducing antibodies that are associated with ARIA. Biogen successfully navigated this issue in its Phase 1b aducanumab program, but it remains difficult to choose the optimal regimen. ARIA events are certainly worrisome, but in most cases are asymptomatic or associated with mild, reversible symptoms. We (and health authorities) are learning that ARIA events are manageable, allowing more aggressive amyloid reduction.
Q: I have not before seen CSF tau data from the aducanumab program—what do you think about them?
A: I think the CSF ptau data are quite encouraging, supporting the idea that reducing amyloid has a beneficial effect on downstream neurodegeneration.
Q: Final word?
A: Overall, this is enormously encouraging for the field. While we need to fully digest this complex data, the results seem to support the amyloid therapeutic hypothesis—very good news for multiple drug development programs underway—and may even indicate that disease-modifying therapy for AD is nearing the clinic.
[Questions by Gabrielle Strobel]
Roche/Genentech
We learned from our early studies with low doses of gantenerumab that exposure is what drives amyloid lowering as well as clinical signals. So we took the time in the OLE to maximize the gantenerumab dose and to develop a way to titrate for E4+ and E4- patients that would minimize interruption (by minimizing ARIA as well as dose reductions), and ensure that almost everyone achieves the highest dose.
We also planned a longer study than most of the AD DM trials (24 months). Based upon recent announcements in the field, and our own work showing that most patients on chronic gantenerumab normalize their amyloid levels, we think that we have optimized our studies to show a strong treatment effect. Recruitment will likely complete in our two ongoing Phase 3 trials ahead of schedule. We look forward to learning in a well-controlled trial whether subcutaneous gantenerumab can transform patient outcomes in AD.
Intra-Cellular Therapies
This is a very interesting development. It seems surprising that the difference between the two studies in the numbers of subjects who had sufficient exposure to the high dose could have accounted for that much difference in the overall results, so I will be interested in seeing more analyses on this point.
On the call, Biogen suggested, as they have done before, that reduction in brain amyloid is a predictor of clinical response. They pointed out that among subjects who met the sufficient treatment criterion, the reductions in brain amyloid were similar in the two trials (Slide 22). However, the reductions in brain amyloid were also similar for the low and high doses comparing both trials overall (Slide 20). Does that weaken the support for a correlation between amyloid reduction and clinical response?
Regarding futility analyses, my approach is to conduct good Phase 2 trials first. But if you are going to go directly from limited data to large Phase 3s, then futility analyses are probably necessary. How and when futility is assessed, and the criteria for declaring futility, are important. For example, it is not clear to me why the futility analysis was done only on the OTC subjects. Regarding pooling, Biogen suggests that at the time of the futility analysis, EMERGE was showing a trend toward efficacy, so if the futility had been assessed in the two trials separately, perhaps the studies would not have been discontinued at that time.
We should keep in mind that these latest analyses are also “interim” in the sense that the trials were never completed. That’s the biggest problem with futility analyses—unless the signal to stop is very clear, you never know what might have happened had you done the trial as originally planned.
University of California, San Diego
These results are definitely newsworthy as they arise from some of the largest, fully powered, and expensive clinical trials that have been undertaken to date with amyloid-lowering treatment. The results however, are now presented post-study discontinuation, and with the presumption that the data were evaluated unblinded, with the inherent risk of introducing bias into these post-trial analyses.
At this point, we are limited in understanding the results. They are presented as percentage change, without inclusion of the actual values of the change scores or the variance terms of the outcome measures, with which to understand the effect sizes.
The results across the two trials appear to be inexplicably divergent, particularly in the light of apparently similar (to identical?) protocols. The lack of converging evidence across trials weakens the strength of the evidence that there is a robust or a transformational treatment with a large effect size at hand; however, any definitive statement must be withheld until more is known.
Applying what is available on a rough percentage change comparison basis on the CDR Sum of Boxes (CDR SOB), the results do not appear to achieve the same very encouraging magnitude of effect size of the initial Phase 1B trial (Sevigny et al., 2016). When data are presented to allow comparison of standardized effect sizes, it will be important to cross-compare these with other intervention types, including findings of LipiDiDiet (Soininen et al., 2017), which achieved significant results on the CDR SOB.
Of more general trial methodological concern is the haste with which experimental treatments including monoclonal antibodies move from very early phase to fully powered late-stage Phase 3 RCTs, only to learn later that the doses being tested are insufficient, that higher doses are needed, and that there are key responder subgroups that are under-recognized. More focus on Phase 2 proof of concept seems well warranted.
References:
Sevigny J, Chiao P, Bussière T, Weinreb PH, Williams L, Maier M, Dunstan R, Salloway S, Chen T, Ling Y, O'Gorman J, Qian F, Arastu M, Li M, Chollate S, Brennan MS, Quintero-Monzon O, Scannevin RH, Arnold HM, Engber T, Rhodes K, Ferrero J, Hang Y, Mikulskis A, Grimm J, Hock C, Nitsch RM, Sandrock A. The antibody aducanumab reduces Aβ plaques in Alzheimer's disease. Nature. 2016 Aug 31;537(7618):50-6. PubMed.
Soininen H, Solomon A, Visser PJ, Hendrix SB, Blennow K, Kivipelto M, Hartmann T, LipiDiDiet clinical study group. 24-month intervention with a specific multinutrient in people with prodromal Alzheimer's disease (LipiDiDiet): a randomised, double-blind, controlled trial. Lancet Neurol. 2017 Dec;16(12):965-975. Epub 2017 Oct 30 PubMed.
University of Gothenburg
This new aducanumab analysis is big and welcome news in Sweden, as well. With the proviso that there seems to be a certain amount of confusion regarding the futility analysis, the presented data appear appropriate and make sense.
I am hopeful that this may unlock our field, and that we now will learn more about what a meaningful biomarker change might look like in AD. In particular, I am hoping that we will soon see plasma NfL and P-tau data from these studies.
Washington University School of Medicine
Q: What are your thoughts about the wisdom of futility analyses in AD trials, given what our field is learning about exposure and response times in AD treatment?
Futility analyses are complex and depend on the specific circumstances and how they are designed. If the cutoff is too conservative, then a futility analysis adds little advantage to adapt or change direction, but if too liberal, it can derail understanding the outcomes of a potentially positive trial.
After taking years for a program to get to Phase 3 trials, I prefer to see more adaptive actions based on a futility, rather than complete shutdown of a program. The design factors of futility analyses are tough decisions that have to be made without full knowledge of outcomes. Thoughtful and wise people have been tasked with these decisions, but there is no clear universal approach that’s best.
This topline data shows that a subset of ENGAGE patients who got uninterrupted 10mg/kg dose had a significant reduction of PET amyloid and also had improved functioning on CDR-SB, with ARIA and headaches the most common side effects.
It is great for the field that Biogen is pursuing these analyses instead of giving up. If the results are truly internally consistent across multiple domains and biomarkers, then this is a breakthrough for the field. We will need to wait and see more data at CTAD in December.
University of Melbourne
This is good news for the AD field in general and for the Aβ theory in particular. When Biogen announced its futility results in March, I thought that something was wrong, based on its earlier data. To read in its announcement now that “the result of the futility analysis was incorrect” vindicates the scientific validity of Sevigny et al., 2016.
The higher dosage in the combined dataset is clearly showing efficacy, and this we should celebrate irrespective of whether Biogen achieve registration. These results breathe new life into this therapeutic strategy, notwithstanding the incidence of ARIA-E.
Of note, gantenerumab, BAN2401, solanezumab, and donanemab are all still in trials. I hope they will all be successful.
The next big step for the field will be to identify a safe low dose for the BACE inhibitors, to be administered at the earliest possible preclinical stage, and to use them in combination with a passive immunotherapy.
References:
Sevigny J, Chiao P, Bussière T, Weinreb PH, Williams L, Maier M, Dunstan R, Salloway S, Chen T, Ling Y, O'Gorman J, Qian F, Arastu M, Li M, Chollate S, Brennan MS, Quintero-Monzon O, Scannevin RH, Arnold HM, Engber T, Rhodes K, Ferrero J, Hang Y, Mikulskis A, Grimm J, Hock C, Nitsch RM, Sandrock A. The antibody aducanumab reduces Aβ plaques in Alzheimer's disease. Nature. 2016 Aug 31;537(7618):50-6. PubMed.
University of California, San Diego
These trial reanalyses look promising and have certainly rekindled enthusiasm for the amyloid hypothesis.
I am a site investigator in one of these Biogen trials, hence will only comment in general terms because there may be efforts to re-enroll subjects. Firstly, futility is a useful addition to trials, to prevent exposing people to unnecessary treatment. The decision to have pooled data across both trials for the aducanumab futility analysis seems reasonable. From the data presented, there were fewer people exposed to high dose for as long as the other dose groups at the time of the futility analyses, though without seeing the analyses in great detail, it is hard to assess the impact.
We do not know what the cognitive or functional trajectory should look like after amyloid is removed from the brain. Brain recovery might continue after plaques are cleared, so there may be a lag in clinical outcomes versus treatment. From bapineuzumab and active immunization studies with autopsy follow-up, at least some patients in whom amyloid was cleared had continued disease progression and extensive tau pathology. It seems there is sufficient uncertainty about amyloid removal to give pause to incorporating futility analyses of clinical outcome measures. Perhaps futility could be conceived in terms of whether the treatment successfully removes amyloid or not, and clinical measures need to be studied for some time longer, such as three to six months longer, after amyloid has been removed.
With a larger N, small clinical effects can become significant. Interestingly, while the study designs were comparable, there are significant effects on CDR-sb, ADAS-cog, and ADCS-ADL-MCI in EMERGE but not ENGAGE—although ENGAGE did show trends in the same direction in the high-dose ITT analysis. The dose effects on biomarkers are incompletely presented. It would be helpful to know how effectively lower doses led to changes in amyloid PET over time compared with the 10 mg/kg subset.
And it would be helpful to see baseline and change data for all clinical outcome measures in the two trials to understand longitudinal change in the placebo groups and treatment-placebo differences. It is difficult to tell from the data presented for EMERGE and ENGAGE what a 23 percent difference in CDR-sb in high-dose versus placebo translates to in terms of absolute points. There is no universal agreement about what effect sizes might be clinically meaningful in AD trials. A recent paper in Alzheimer’s and Dementia looked at cognitive, functional and CDR-sb data. Clinically meaningful change for the CDR-sb was estimated to be about 1 point for MCI/mild AD in this analysis of multicenter data (see the figure in Andrews et al., 2019).
References:
Andrews JS, Desai U, Kirson NY, Zichlin ML, Ball DE, Matthews BR. Disease severity and minimal clinically important differences in clinical outcome assessments for Alzheimer's disease clinical trials. Alzheimers Dement (N Y). 2019;5:354-363. Epub 2019 Aug 2 PubMed.
Acumen Pharmaceuticals
I have consulted for Biogen Inc. as well as a variety of other pharmaceutical companies.
The Biogen announcement on October 22 indicating that they would submit the data from the EMERGE and ENGAGE trials to FDA for possible regulatory approval was unexpected but welcome news for AD research. The regulatory deliberations will be based on very detailed analyses of the two studies, both individually and taken together, and what the judgement of the FDA along with an advisory committee might be remains speculative.
Perhaps more important are the scientific implications of the Biogen data. While the two studies were initially determined to be futile, the availability of additional data apparently convinced Biogen that a signal was present, at least in some individuals. That conclusion should be put in context of findings from other monoclonal antibodies.
The solanezumab EXPEDITION studies showed consistent but small effects on cognitive outcomes, with the smallest effect seen in the final EXPEDITION-3 study. (The A4 study, with a 4-fold higher dose in participants with preclinical AD should finish in 2022, and the DIAN study should be reported next year). The BAN2401 Phase 2 study showed 47%, 30%, and 26% slowing of progression on the ADAS-cog, ADCOMS and CDR-SB respectively.
The EMERGE study showed 23% reduction of progression using the CDR-SB and 27% and 40% slowing of progression using the ADAS-cog13 and ADCS-ADL-MCI respectively. The ENGAGE study showed no effect on the CDR-SB, and only trends for slowing of the ADAS-cog13 (12%) and ADCS-ADL-MCI (18%). The reasons for this disparity are being studied.
These intriguingly similar results, using 3 different monoclonal antibodies with different binding epitopes, increase the weight of evidence that targeting Aβ species or amyloid can change the progression of AD. While the studies individually are of great interest, their value to the field could be enhanced substantially by analyses of the combined data. Regardless of the ultimate regulatory outcome of the aducanumab submission, sharing and additional analyses of these extremely well characterized datasets would be of great scientific value.
Certara
The sudden discovery of a beneficial effect in one, but not in a second Phase 3 outcome for aducanumab could be due to a combination of a genuine dose-dependent effect and a statistical effect due to an imbalance of a responder genotype.
We applied our Quantitative Systems Pharmacology analysis (Geerts et al., 2018), basically a biophysically realistic computer model of humanized neuronal circuits, based on the non-linear neurobiology of short and long Aβ isoforms with the pharmacodynamic interaction of three common genotype variants and with AChE-I. Virtual patient simulations suggest that the clinical benefit on a cognitive readout is indeed exposure-dependent but that an additional imbalance of one genotype is needed to reach the 20 percent improvement over placebo (poster to be presented at CTAD). The model predicts also that part of the beneficial effect is driven by the placebo in the trial. It would be great to test these outcomes with the actual analysis, as this might help to better understand the role of amyloid and position the therapy for the right patient population and condition.
References:
Geerts H, Spiros A, Roberts P. Impact of amyloid-beta changes on cognitive outcomes in Alzheimer's disease: analysis of clinical trials using a quantitative systems pharmacology model. Alzheimers Res Ther. 2018 Feb 2;10(1):14. PubMed.
Arizona Alzheimer's Consortium
In a commentary accompanying the 2016 PRIME Nature paper (Reiman, 2016; Sevigny et al., 2016), I expressed excitement about aducanumab’s ability to reduce Aβ plaque burden and caution in drawing conclusions about its cognitive benefit until Phase 3 trial results were in. I noted that if the Aβ reduction were associated with a cognitive benefit in those trials, it would be a game-changer. We wish the planned trials had been completed such that we’d have a more definitive answer, but even so, the field may be closer to that game-changing moment than before.
While we are eager to review the findings when they are described in more detail, Biogen’s announcement suggesting a cognitive benefit in those Alzheimer’s patients treated at the highest dose for a sufficiently long duration is a timely shot in the arm for the field. If the Aβ plaque reduction by aducanumab and/or other drugs now in development is followed by a clinical benefit, this data would profoundly affect the fight against AD. Here’s why:
My API and A4 colleagues and I proposed and generated initial NIA grant support for an initial 24-month prevention trial in cognitively unimpaired Aβ+ persons. We suggested that if a plaque-reducing treatment was associated with a clinical benefit in ongoing trials—alone, or better yet in combination with an attenuation in downstream tau and neurodegeneration biomarkers—it might be possible to demonstrate that the treatment would have a similar impact on those “theragnostic” biomarkers in unimpaired persons, qualify those biomarkers for use as “reasonably likely surrogate endpoints” in prevention trials, and support accelerated approval of a prevention therapy before 2025. There is a lot more to learn about the nature of and relationship between a treatment’s different biomarker (including blood-based biomarker) and clinical effects. That said, we are eager to give this possibility the best possible shot.
In light of the potentially premature death and subsequent resurrection of aducanumab, I’m reminded of two things. One, there’s peril in conducting futility analyses and making discontinuation decisions based on limited data and assumptions that haven’t been tested with clinically proven treatments. This includes assumptions related to the temporal relationship between a treatment’s biomarker and its clinical effects. While I am very encouraged by the trial findings that I have been able to review so far, we need to do our best to avoid premature discontinuation decisions. We have so much to learn from completed trials, whether or not they are successful. Two, we are lucky Biogen continued to vet the findings in a rigorous and timely fashion. I salute our industry colleagues who stay in the game, despite past failures, to address this problem in the most fundamental way.
We still have a fighting chance to improve the lives or our patients and families, and of finding a prevention therapy that works within the next five years. Let’s do everything we can to make that happen.
References:
Reiman EM. Alzheimer's disease: Attack on amyloid-β protein. Nature. 2016 Aug 31;537(7618):36-7. PubMed.
Sevigny J, Chiao P, Bussière T, Weinreb PH, Williams L, Maier M, Dunstan R, Salloway S, Chen T, Ling Y, O'Gorman J, Qian F, Arastu M, Li M, Chollate S, Brennan MS, Quintero-Monzon O, Scannevin RH, Arnold HM, Engber T, Rhodes K, Ferrero J, Hang Y, Mikulskis A, Grimm J, Hock C, Nitsch RM, Sandrock A. The antibody aducanumab reduces Aβ plaques in Alzheimer's disease. Nature. 2016 Aug 31;537(7618):50-6. PubMed.
Harvard Medical School, Massachusetts General Hospital
Our laboratory has a long-standing collaboration with Biogen on the preclinical studies of aducanumab. We used an animal model of amyloidosis, old Tg2576 mice, to show that aducanumab crossed the blood-brain barrier, engaged the target, and cleared amyloid.
This particular mouse model exhibits intraneuronal calcium elevations due to disruptions in calcium homeostasis. Thus, we used calcium dynamics as a functional indicator of treatment efficacy. Indeed, aducanumab restored neuronal calcium homeostasis, likely restoring neuronal network functions underlying cognitive deficits.
Interestingly, 10 mg/kg was the efficacious dose required to do so. Studies on the mechanisms of aducanumab action are currently ongoing in our laboratory.
References:
Kastanenka KV, Bussiere T, Shakerdge N, Qian F, Weinreb PH, Rhodes K, Bacskai BJ. Immunotherapy with Aducanumab Restores Calcium Homeostasis in Tg2576 Mice. J Neurosci. 2016 Dec 14;36(50):12549-12558. Epub 2016 Nov 3 PubMed.
Chiesi Farmaceutici S.p.A.
The last week saw a debate on potential reasons to explain the divergent outcome on the primary variable (CDR-SB) emerging from the new analyses of the two interrupted (apparently identical) Phase 3 studies on aducanumab. Compared with the placebo group, the high-dose aducanumab group had a 23 percent reduction in the mean CDR-SB decline in the EMERGE study but a 6 percent worsening in the ENGAGE study. Biogen explained that since the ENGAGE study started a month before the EMERGE study, fewer patients were exposed to the higher dose (allowed with two protocol amendments) and those who did get the high dose were exposed for a shorter duration of time compared to the EMERGE study.
Available data on the two studies are limited, especially regarding the absolute extent of the CDR-SB decline in the different treatment groups and their clinical relevance. We do not know the number of dropouts in the different treatment arms, the reasons for dropping out and, importantly, the disease severity of the patients abandoning the study in the placebo and the aducanumab-treated groups. We will learn much more at the next scientific meetings.
In the meantime, I noticed that in the so-called Opportunity to Complete (OTC) population of the Larger Dataset analysis, the EMERGE study has 102 patients fewer than the ENGAGE study. This may be explained by the fact that the EMERGE study started one month after the ENGAGE study. However, this difference in patients is restricted to the placebo (-56 patients) and the aducanumab low dose (-41 patients) groups (see table below). As expected, in the Intention-to-Treat population, there were no significant differences in number of patients between the two studies in either group. The reasons for the selective loss of patients in the placebo and the low-dose groups in the EMERGE study are unclear; higher dropout rates in the placebo and low-dose groups appear unlikely.
BarcelonaBeta Brain Research Center, UPF & Alzheimer's research and other cognitive disorders unit, Clinic Hospital
After listening to Biogen announce that it will file to the FDA based on the results of its recent trials suggesting a cognitive benefit in those patients treated at the highest dose, I share the cautious excitement that many of my colleagues express. It is true that we are looking forward to seeing and digesting the data carefully, but nevertheless this is good news that benefits the Alzheimer’s field.
Filing the data means, as a start, that there is sufficient signal for considering a potential approval. I hope that this fact, independent of the FDA’s final decision, will result in a renewed interest of companies and investors in the field. In addition, if the result holds, it is telling us that, even at a stage where the brain is not only full of deposited amyloid but also where tau tangles have spread beyond the middle temporal lobe, removing amyloid has an impact at different levels, showing a downstream signal on soluble tau biomarkers and a clinical benefit.
The emerging question for me would be, what will be the benefit of removing amyloid earlier in the continuum, before tau tangles spread beyond the middle temporal lobe? I belong to the group who believe that our chances for defeating Alzheimer’s are higher if we go earlier. I am fully aware of the complexities of designing secondary prevention trials: finding the right population, defining primary endpoints, powering the trial (one problem there is that we tend to use data from observational studies which do not account for the placebo effect), attrition rates, to mention a few.
Nevertheless, a group of us have been already working on creating research infrastructures that can accommodate all kind of initiatives for the prevention of AD. We have built registries, some of them targeting specific populations that will suit secondary prevention trials, longitudinal cohorts that populate our own studies, and even designed platform trials for the secondary prevention of AD.
I am personally directing a research program fully focused on Alzheimer’s prevention, strongly involved in TriBEKa, EPAD, and AMYPAD, so cannot deny a strong bias toward Alzheimer’s prevention, however, this does not preclude me from encouraging companies that have the right compound to move into secondary prevention. It is probably timely to do it now, although I would also dedicate sufficient time to designing a trial that can overcome the hurdles and complexities of secondary prevention.
I do remain optimistic and hope that others are, and that we can keep on working together to defeat Alzheimer’s disease.
Make a Comment
To make a comment you must login or register.