This is Part 2 of a two-part series. See also Part 1.
8 August 2012. At the Alzheimer’s Association International Conference, held 14-19 July 2012, leaders of three ongoing preclinical treatment initiatives answered questions from a packed audience and from moderator Howard Feldman of the University of British Columbia in AAIC’s Canadian host city of Vancouver. Panelists were Paul Aisen of the University of California, San Diego; Reisa Sperling of Harvard Medical School; Randy Bateman of Washington University, St. Louis, Missouri; and Eric Reiman, Pierre Tariot, and Jessica Langbaum of the Banner Alzheimer’s Institute in Phoenix, Arizona.
Feldman: You have shown us a variety of different trial designs and sample sizes. Tell us why you think your design is best and what a reasonable effect size is.
Sperling: For A4 we target a 25 to 30 percent effect size, partly because that is what is known from previous trials. We would love to achieve 50 percent but have not seen that yet in AD therapeutic studies. We need better drugs for that. We chose a large group size because we do not know the exact rate of progression in our population, so we built in room for that.
Bateman: We have 40 mutation carriers in each arm, so the comparison is between 40 carriers on drug A versus 40 carriers on placebo. Be aware that we are talking about apples and oranges, because our trials use biomarker effect sizes, and A4 and API use clinical effect sizes.
That said, for example, for brain amyloid deposition, we are not expecting a 25 percent slowing of amyloid accumulation. We expect to stop accumulation or reverse it. Our points of reference are the published bapineuzumab and gantenerumab effect sizes on amyloid PET and CSF (Rinne et al., 2010; Blennow et al., 2012; Ostrowitzki et al., 2011).
Tariot: The reason so many people are in the room today is that there is a recognition that we are in uncharted space. My point is: This is unprecedented, so we are making assumptions and have to be guided by the data.
Feldman: Each of you will measure cognition in some way, but you all do it a bit differently. Each of you uses a different composite outcome.
Langbaum: The composite endpoints are actually very similar. I think the differences between the API and A4 primary endpoints will be quite minor. We found ours to be most sensitive to presymptomatic longitudinal decline in data from the Colombian Antioquia cohort and the Rush Memory and Aging Cohort and Religious Order Study. We derived two very similar composites from these two cohorts independently. A4 developed their composite using different longitudinal datasets, yet the results were similar to those found by API. In the end, certain cognitive domains are important, and each of the composites taps them in a way that is informative in the asymptomatic period. There is enough overlap between our composites to make the data very comparable.
Audience question: It’s good to bring this together through CAP. How do you address the issue of variability in the CSF measures?
Bateman: I am confident we can use these measures. Anne Fagan handles the DIAN and the API samples, so there is no lab-to-lab variability. Many of the other issues with variability and quality control of imaging and CSF are being addressed internally, with external advisors, and in separate ongoing initiatives. [Editor’s note: See ARF AAIC story.] Incidentally, that is why public funding of this trial is so important. With public funding, we can make samples available to groups that develop and validate new assays.
Audience question: How about ApoE in A4?
Sperling: The goal is to stratify but not pre-specify group sizes according to ApoE genotype.
Audience question: Have you considered combination trials?
Bateman: We gave great consideration to that. Unfortunately, there are operational and practical challenges of giving two drugs, neither of which is approved, much less from the same company. But scientifically and medically, there is a strong rationale for combination treatment.
Sperling: We would love a 2-by-2 design for an anti-amyloid antibody and a BACE1 inhibitor, or later an Aβ antibody and a tau antibody, but we cannot do that yet. That will be the A6 trial.
Audience question: Can DIAN make a pooled placebo work?
Bateman: Yes, because the primary and key secondary outcomes at this stage are the biomarker outcomes.
Tariot: On all the issues we are discussing, we need to keep in mind that there have been informal regulatory discussions but no formal approval yet from the FDA, the Colombian Ministry of Health, or ethics committees. This is new territory and new issues may come up.
Audience question: Does A4 plan to enrich for family history or subjective memory complaint?
Sperling: We will advertise for individuals who have a family history. We will admit people who report a subjective memory complaint but perform normally on the cognitive instruments.
Audience question: How about people younger than 70 who have a positive amyloid scan?
Sperling: We have discussed this. We know there are people in other studies now who are positive for amyloid. The reason we chose 70 was to be able to find a workable rate of amyloid positivity during screening, but if you are 69, we want you to be able to get in. We still need the FDA’s okay on that.
Audience question: How about treating earlier?
Aisen: We will be tracking people who were amyloid-negative at screening and seeing what happens to them. They could be in a next trial. We do not have the proper tools to do primary prevention now, but are working toward those.
Audience question: Both DIAN and API take into account that most people do not want to know their status, and the trials are powered accordingly. That is good. But enrolled people may find out their status through side effects. That needs to be said clearly.
Tariot: We will address this through informed consent. Some of the API participants in Colombia may not have much secondary education, but they are very functional and highly perceptive. They understand safety and efficacy. They are very practical.
Audience question: How are you thinking about risk-benefit? About death, severe morbidity?
Reiman: Based partly on the feedback we got from family members and external advisors, we sought an investigational agent that had evidence of target engagement (in this case, a potent amyloid-modifying effect and some diversity in the amyloid species that may be targeted) that may be associated with a reduced risk of adverse effects in healthy individuals, and safety and tolerability data (e.g., no reported microhemorrhages in the first 250 patients who have been exposed to the drug). It is important to recognize that the treatment is investigational. There is no guarantee of benefit; there is no certainty about safety or tolerability. We will have more information about safety and tolerability from ongoing trials before this preclinical trial is underway. During the selection process, we and the other committee members discussed the issue of how much efficacy (including biomarker and clinical) data would be needed before selecting a treatment. Those data are not yet available for crenezumab; however, our trial includes a 24-month interim analysis that will help in the decision to continue the trial. On managing potential risks, we believe transparent communication with the families is essential.
Tariot: We have met several hundred of these family members. Some came to our homes in Phoenix last fall. One night, 10 participants in their thirties were at our dinner table. None knows his or her genetic status. At first, some did not know each other, but by the end of dinner they looked at one another and said, “How awful that half of us will die from AD.” In other words, they know their risk. In addition to concerns for risk from the drug, also remember what the risk of AD means to them. All the standard clinical trial safety standards are being built into this trial. But they are a special population.
Bateman: The DIAN families have lived with their mutation for generations. They know what they want and what they do not want. In DIAN, there are family representatives on the steering committee and on the clinical trials committee. The family members have engaged as key partners in the design of the DIAN trial, including genetic status blinding and increasing the percent of participants receiving the active drug. The family representatives will see any toxicities and side effects as the trials unfold.
Audience question: What would be the implications of an entirely negative result of the bapineuzumab and solanezumab programs on your trials?
Aisen: Their Phase 3 programs have statistical analysis plans that give not only overall success or failure, but also information on biomarker engagement. If there is no efficacy but a strong biomarker response, would that dissuade us from using the agent? No. The clinical implications of treatment in preclinical AD are different.
Audience question: What about a negative trial with significant toxicity?
Aisen: We anticipate that the drug we choose will have risk. There needs to be an explication of risk and benefit. Some risk is acceptable. It would be a mistake to demand that people with amyloid who are cognitively normal should only be exposed to totally safe compounds.
Bateman: Paul is exactly right; the data will drive the decision. Of course, there is a balance of risk and benefit. In our case, the participants have certain risk of disease.
Tariot: There have been multiple private and public meetings here and over the last few years. Despite the aspersions some of us may sometimes cast on regulatory agencies, they have been extraordinarily helpful and willing to suggest to us what would be acceptable. Without that, our own study would not have been able to go forward.
Feldman: Clearly, the FDA wants to be partners because they see the benefit to the public. Sometimes we scientists hide behind them when, in fact, the tough decisions are in front of us.—Gabrielle Strobel.
This is Part 2 of a two-part series. See also Part 1.