2012 Autumn: Options and Methods to Improve Cognitive Assessment in Clinical Trials of AD and Its Precursors Workshop

Co-chairs: Holly Posner, MD; Phil Harvey, PhD

Objectives:

Understand the challenges in assessment of cognitive enhancement studies in AD and its precursors.  Critically evaluate whether sophisticated statistical techniques are adequate to overcome psychometric challenges in existing instruments. Propose alternatives in the event that it appears as if current data sets cannot be salvaged. Evaluate new approaches to assessing outcome using new performance-based  strategies.  Participants will include Terry Goldberg, PhD, (LIJ Hofstra School of Medicine),  Dan Mungas, PhD, (UCI Davis), Yaakov Stern, PhD, (Columbia University), Veronica Logovinsky (Eisai), and Allitia DiBernardo, JnJ.

Background: The Alzheimer’s Disease Assessment Scale (ADAS-cog) is a composite of several neuropsychological tests and is the de facto cognitive outcome standard in AD trials.  This scale was rationally derived on the basis of knowledge and measurement techniques that are now close to 30 years old and many of the items lack sensitivity to a broad range of impairments in the AD spectrum.  Thus, the instrument typically generates data that is insensitive to impairment in milder cases and hence is handicapped in its ability to detect changes.

Multiple pharmaceutical companies have re-analyzed ADAS-cog and other AD/ MCI/ Prodromal AD endpoints in an effort to improve the sensitivity of the outcome measure to detect improvement from a pharmaceutical or biologic intervention. Some companies have used purely statistical methods and others have also used psychometric analysis, like IRT. Most have also attempted to combine multiple outcomes measures, such as MMSE, CDR, and other scales along with ADAS-Cog data. A sampling of these efforts will be presented.

There is an understandable sense of urgency regarding solutions to the low sensitivity issue and an understandable reluctance to wait for alternative instrumentation to be developed. Since very large and expensive drug development decisions can rest on the quality of outcomes assessments, key stakeholders require a thorough understanding of whether statistical techniques can overcome the intrinsic limitations of existing assessment methods.  At the same time, if statistics cannot resurrect data collected with older methods, then new strategies must be developed or alternative regulatory approaches will be needed.  Critical evaluation of the competing strategies of “charge forward now” vs. “develop something that works” is required. Further there are recent developments in co-primary assessments in other conditions, such as schizophrenia, that have been applied with considerable success in mildly impaired cases.

Potential deliverable: we plan on producing a white paper, starting out by laying out the issues:

·         Questionable sensitivity

·         Possible inadequate coverage

·         Urgent need to assess less impaired patients

·         The  tension of the need to run trials vs. the need to have an outcome measure that work.

This would include a review of the publicly available information on salvaging  ADAS-cog data with statistics.  It would also outline previous efforts at consensus-based outcomes development, including evaluation of the time-line for this.  Finally, the white paper could comment on the regulatory perspective and whether some thinking, like we have seen in schizophrenia, on performance based measures of everyday functioning could be a substitute for current cognitive measures supplemented by clinical judgments and informant reports.