Posted 1 March 2005

Report by Marilyn S. Albert, PhD and Michela Gallagher, PhD; Johns Hopkins University, Baltimore, MD

Introduction
This report summarizes the discussion and recommendations of a conference entitled "Assessing Cognition for Emerging Therapeutics in Alzheimer's Disease". The goal of the conference was to identify areas of consensus regarding cognitive changes that occur with age and in the earliest stages of Alzheimer's disease, the neurobiological changes most responsible for these cognitive changes, and unresolved issues that require future investigation. The conference occurred on September 13-14, 2004, at the Baltimore Waterfront Marriott Hotel, Baltimore, MD. It was co-sponsored by the Alzheimer Research Consortium, Johns Hopkins University and the Johns Hopkins Alzheimer's Disease Research Center. Drs. Albert and Gallagher organized the conference and served as co-chairs. This report has been reviewed by some, but not all, of the participants in the meeting. Additional comments and details from meeting attendees are particularly welcome.

Participants:
Session 1
Marilyn Albert, Johns Hopkins University;
David Bennett, Rush-Presbyterian, Chicago;
Steven Ferris, New York University;
Richard Mohs, Lily Pharmaceuticals;
Ron Petersen, Mayo Clinic;
Barbara Sahakian, Cambridge University;
Peter Snyder, Pfizer Pharmaceuticals;
Yaakov Stern, Columbia University, NY.
Session 1 - Moderators
Dennis Selkoe, Brigham & Women's Hospital, Harvard Medical School;
Hugh Hendrie, University of Indiana

Session 2
Carol Barnes, University of Arizona, Tucson;
Mark Baxter, Oxford University;
Charles Duffy, University of Rochester;
Michela Gallagher, Johns Hopkins;
Mark Gluck, Rutgers University;
Mark Moss, Boston University;
Peter Rapp, Mt Sinai Medical School, NY.
Session 2 - Moderators
Brad Hyman, Massachusetts General Hospital, Harvard Medical School;
Marilyn Albert, Johns Hopkins University.

Session 3
Karen Hsiao Ashe, University of Minnesota;
Brad Hyman, MGH; Frank LaFerla, UC Irvine;
David Morgan, University of South Florida;
Richard Morris, University of Edinburgh;
Lennart Mucke, UCSF; Don Price, Johns Hopkins.
Session 3 - Moderators
Michela Gallagher, Johns Hopkins University;
Peter Rapp, Mt Sinai Medical School.

1. Cognitive Change in Humans in the Early Stage of Alzheimer's Disease

1a. Cognitive Change in the Early Stage of Alzheimer's Disease
Much is known about the cognitive changes seen in the earliest stages of Alzheimer's disease (AD). This knowledge comes from studies of patients with mild AD as well as from examinations of subjects who are symptomatic, but not yet demented. The latter group is commonly referred to as having mild cognitive impairment (MCI). Consistent findings from these studies has led to the consensus that the earliest stage of AD is characterized by an impairment in episodic memory (i.e., learning and retention of new information). This impairment can be demonstrated on tests of both verbal and non-verbal memory, using both standardized paper and pencil tests as well as computerized tests of memory. The tests that are most useful for demonstrating this episodic memory impairment in symptomatic individuals all share the common feature that they assess both immediate learning and retention over a delay. To date, there is insufficient evidence to determine whether one particular episodic memory test is better than all the others at discriminating individuals in the earliest phase of AD. Likewise, there is a consensus that early identification of patients with AD is assisted by measurement of at least one other cognitive domain, above and beyond memory. However, there is no consensus regarding which additional domain is most discriminating. Some investigators report that changes in executive function (as assessed by tests that emphasize set formation and set shifting) are most helpful, while others report that alterations in semantic memory (e.g., naming) is important to assess.

1b. Neuropathological Changes Underlying the Earliest Cognitive Alterations in AD
Neuropathological studies of individuals in various stages of AD pathology have led to the consensus that the entorhinal cortex (particularly layers II and IV) demonstrates the most profound neuronal loss in the earliest stage of AD, primarily because of the accumulation of neurofibrillary tangles and neuritic plaques. The primary efferent pathway from the entorhinal cortex goes to the hippocampus, and this region appears to be the next major area of pathological involvement in AD (particularly the anterior hippocampus). Based on studies of brain lesions in humans and non-primates, it is known that these medial temporal lobe regions are critical for normal episodic memory. Thus, a consensus has emerged that the gradual accumulation of AD pathology in medial temporal lobe structures, particularly in the entorhinal cortex and hippocampus, is responsible for the gradual development of an episodic memory impairment that begins in the prodromal phase of AD and continues once patients have established dementia. There is insufficient data to reach a consensus on the nature of cognitive decline in normals that differentiates individuals who will progress to MCI from those who will not or, by extension, what aspect of AD pathology may be responsible for these alterations (e.g., accumulation of soluble amyloid beta peptide).

1c. Non-invasive methods of measuring the earliest cognitive changes associated with AD
Non-invasive imaging methods have been used to measure alterations in brain structure and function among individuals in the earliest stages of AD. Magnetic resonance imaging (MRI), and positron emission tomography (PET) are the most widely used methods. MRI methods have focused primarily, though not exclusively, on measuring the volume of regions of interest within the medial temporal lobe (e.g., the entorhinal cortex and hippocampus) or the brain as a whole. There is a consensus that these measures of atrophy are an indirect reflection of neuronal loss secondary to AD pathology, and that such measures correlate with performance on episodic memory tests in MCI and AD cases. Functional imaging methods that use a radiotracer (i.e., PET and SPECT), also show alterations in brain function early in the course of AD. However, when voxel-based methods of analysis are employed the brain regions showing the greatest alteration are not within the medial temporal lobe, but rather in the posterior cingulate and the temporoparietal cortex. These functional changes also correlate with cognition in MCI and AD cases. It has been hypothesized that the differences in brain regions identified by these two types of methods derive from their differential sensitivity to alterations within the medial temporal lobe or to terminal projections from the medial temporal lobe; a consensus has not yet emerged about this issue. Other non-invasive imaging methods that have been employed include functional MRI and MR spectroscopy. As these methodologies have been less widely used, there is less consensus about the nature of the changes observed or their relationship to cognition. Likewise, measures from cerebrospinal fluid (CSF) have shown promise but a consensus about their applicability has not been reached. Other more novel methods, such imaging amyloid directly with a radiotracer, are in the early stages of examination in MCI and AD.

1d. Assessing cognition in clinical trials of MCI and ADB
The FDA has recently made it clear that it will continue to require both cognitive testing and a global assessment of function in all clinical trials of MCI and AD. The performance-based cognitive measures must be valid and reliable and can be a single measure or a composite measure. The global clinical assessment should, by definition, be a fairly course measure, in order to assure that any change that is seen is clinically meaningful. In addition, the FDA has recently stated that non-invasive imaging measures that showed differential change in response to treatment would be helpful evidence in the approval of any future medications. However, in order for a non-invasive measure, such as imaging or a biomarker from CSF, to be considered a 'surrogate measure' one would need to show that the degree of change in the surrogate measure was highly correlated with the response to treatment and to the underlying severity of disease. Based on the findings noted in sections 1a-1c above, there was a consensus that measures of episodic memory and at least one other cognitive domain should be included in future clinical trials of MCI and AD, and that non-invasive imaging measures, such as MRI, should be incorporated as well.

Recent clinical trials in MCI have, however, been challenging, primarily because most had a lower than expected conversion rate to AD, and thus there was insufficient power to see a medication effect. Considerable work, therefore, needs to be done to determine how to enrich a sample of MCI cases with those most likely to convert to AD within the study period. As a result, there was not a consensus about whether future trials should include MCI cases or focus primarily on AD. Any resolution about this issue also depends upon an improved understanding of when, in the course of disease, the brain pathology is likely to be most responsive to treatment.

2. Cognitive change with age in the absence of AD

2a. Cognitive change with age in rodents
Much is known about cognitive change with age in rats. Since rats do not normally develop the pathology of AD, these findings are relevant to age-related change in the absence of AD. There is a consensus that the optimal way to study age-related change is to include only optimally healthy animals in analyses, making it possible to study aging in the absence of disease (systemic as well as brain disease). Studies that have used such an approach have shown that there is an age-related impairment in episodic memory that begins to be apparent in middle-aged animals. However, there is considerable variability with advancing age, with some animals clearly memory-impaired and some unimpaired, in comparison to younger animals. The application of unbiased stereological techniques to assess neuronal number in these animals has demonstrated minimal neuronal loss with age in a variety of medial temporal lobe regions and no relationship between degree of memory impairment and neuronal number in these areas. However, measures of connectional integrity (e.g., number of synapses and strength of synaptic connections) and functional plasticity (e.g., LTP and LTD) are altered with age, and these alterations are found in the animals that are memory impaired, but not in those without memory impairments. Thus, there is a consensus that age related change in rodents occurs in the absence of neuronal loss, that it is not seen in all older animals uniformly, and that it results from alterations in the connectivity and responsivity of medial temporal lobe structures and their interconnections with other brain regions.

2b. Cognitive change with age in non-human primates
Studies in non-human primates are quite parallel to those in rodents. Recent studies in optimally healthy rhesus monkeys have shown age-related memory impairments. Unbiased stereological techniques in such animals have shown minimal neuronal loss in the hippocampus and in most cortical regions studied to date. As with rodents, difficult memory tasks show impairments among animals that are middle aged, but performance is highly variable among older monkeys, with some age-impaired and some age-unimpaired, in comparison with younger animals. Thus, there is a consensus that age related change in monkeys occurs in the absence of neuronal loss, and that these changes are not seen in all older animals uniformly. There is converging evidence that alterations in synaptic function and connectivity play a major role in age-related cognitive change in non-human primates, in addition to neurotransmitter alterations related to the dopaminergic and cholinergic systems.

There was a consensus that non-human primates offer a number of unexplored opportunities for the understanding of AD. Several examples were discussed at the meeting, including: the need to examine brain tissue from behaviorally characterized animals with modern molecular techniques (looking at the nature of amyloid deposits, dodecamers, etc), and the delivery of transgenes, RNAi silencers or enzyme inhibitors to selective brain regions (e.g., hippocampus).

3. Cognitive change in transgenic mouse models of Alzheimer's disease
Numerous mouse models of AD have now been developed, and the cognitive deficits in these models have been examined with a variety of tasks. In most of these models, the cognitive deficit emerges with advancing age. There was a consensus that the adequacy of an animal model should be judged, not only by the nature of the brain pathology produced, but also by the fact that cognitive deficits worsen over the life-span, rather than to be present at birth.

All investigators that assess cognition in transgenic mice have focused on characterizing the memory deficit and the one task that appears to be in universal use across laboratories is the Morris Water Maze. This task appears to be sensitive to treatment effects (e.g., antibody treatment clears A-beta and spatial memory improves in response to treatment). There was a consensus, however, that transgenic models need to be evaluated with a range of tasks to clarify the nature of the behavioral response (e.g., antibody treatment in some models improves performance on the Water Maze but does not improve behavior on an inhibitory avoidance task). The inclusion of executive function tasks was specifically mentioned as potentially valuable for characterizing the nature of the cognitive deficits in transgenic mice.