Seeing with new eyes: Finding a path to early intervention trials in Alzheimer's disease

Abstract

The clinical course of Alzheimer's Disease (AD) and other degenerative disorders affecting cognition can be visualized as a progression from normal cognition through the syndrome of Mild Cognitive Impairment (MCI) to dementia. The use of biomarker data can supplement clinical characterization and identification of MCI and dementia pathologies. Clinical staging algorithms that use both clinical and biomarker information can assist in the early identification of AD patients. A comprehensive outcome measure such as the Clinical Dementia Rating Sum of Boxes (CDR-SB), which has components that assess both cognitive and functional domains in parallel deserves consideration as a primary outcome measure for early AD clinical trials.

“The real voyage of discovery consists not in seeking new landscapes but in having new eyes.” – Marcel Proust

It is now recognized that Alzheimer’s disease does not start the day a clinician pronounces the patient as being “demented”, but is the end result of a pathological process that begins years, if not decades before. The progression from asymptomatic pathology through prodromal AD to symptomatic and terminal disease can be viewed as a continuum rather than a saltatory progression (1). Grappling with this continuum in a way that can support meaningful development of new therapies to treat symptoms or perhaps eventually retard or forestall progression of disease manifestations presents us with a set of unique challenges. The recognition that the syndromes of “Dementia” (2) and its precursor, “Mild Cognitive Impairment, or “MCI” (3) are varied in their manifestations, reflecting varying pathological substrates (4, 5), is forcing us to develop new diagnostic tools that hopefully will enable us to tailor our new therapies to specific disease mechanisms (6). More specific biological tools (measurement of CSF A (7, 8), amyoid PET imaging (9, 10), dopamine transporter [DAT] imaging (11), serum progranulin levels (12), CSF synuclein measurements (13), etc) as well as improved clinical and neuropsychological characterization of various MCI and dementia syndromes are becoming available (Table 1). In concert with these newer biological tools is the recognition that amnestic and multi-domain patterns of MCI, convey greater likelihood of progression to an AD-type dementia, whereas non-amnestic MCI with predominant impairment of executive function portends progression to non-AD dementias (6).

Table 1.

Common Causes of MCI and Dementia Syndromes and Related Emerging Biomarkers

Syndrome	Cognitive Features	Potential Biomarkersa
Alzheimer’s Disease	• Amnestic cognitive impairment;	• CSF A
	• Executive dysfunction	• Amyloid PET imaging (PIB, others)
	• Impairment in other cognitive domains (e.g., apraxia,	• CSF tau/phospho-tau
	aphasia, behavioral disturbances)	• Metabolic / functional imaging (FDG-PET, fMRI; • Structural imaging (MRI)
Vascular Dementia	• Executive Dysfunction > Memory impairment	• Vascular “Risk Factors”
	• Focal neurological signs	• MRI White Matter Hyperintensities
Dementia with Lewy Bodies	• Fluctuating attention/awareness	• Dopamine Transporter SPECT imaging
	• Hallucinations/delusions	• CSF a-synuclein(?)
	• Sleep disorder
	• Eventual appearance of extrapyramidal signs
Frontotemporal Dementias	• Behavioral abnormalities	• Regional atrophy on MRI
	• Progressive Aphasia syndromes	• Frontal Hypometabolism (FDG-PET)
	• Executive dysfunction	• TDP-43, progranulin genotype • Decreased Serum Progranulin

^a.See references in text

One problem we face is identifying the point in time at which the various dementia syndromes become suitably recognizable in order to introduce targeted therapies. Alternative diagnostic constructs, such as the “Dubois criteria” (14), that combine clinical syndromic characterization (episodic memory deficit) with diagnostic biomarker data appear to be an improvement over clinical characterization alone. However, the “Dubois Criteria” remain to be fully validated and operationalized in a practical diagnostic system.

Figure 1.

The one-year mean change from baseline for ADAS-Cog and CDR-SOB (1A), ADAS-Cog and FAQ (1B), and FAQ and. CDR-SOB (1C) in enriched vs. non-enriched MCI populations. Clinical and CSF biomarker data for MCI subjects in the ADNI database (accessed August 18, 2009) was used to identify 2 populations of subjects: an “enriched” (red dots) and “non-enriched” (blue dots). 2500 trial simulations were then run to compare outcomes on standard clinical measures in the two groups. Each simulation generates two plotted values on each graph; one red dot representing mean changes from baseline for two tests in the “enriched” group, and one blue dot representing the corresponding mean change pair for the “non-enriched group”.. The enrichment criteria used produced populations of patients that differed greatly in their progression rates, as demonstrated by the near complete segregation of blue and red dots.

Developing a new vocabulary to describe the earliest stages of AD is an important step in advancing therapies for AD. MCI is currently the earliest recognizable stage of cognitive impairment. The terms “prodromal” (15), “pre-dementia” (16) and even “very mild” (17) have been suggested for the phase of AD in which cognitive impairment first begins to make inroads into changing its victims’ lives and those of their families, co-workers, and friends. However, it may be appropriate to adopt a “staging” system for AD, similar to that used to stage cancer. An example of such a schema, not too dissimilar from those proposed by others (18) is shown in Table 2.

Table 2.

A Clinical and Biomarker Staging Scheme for Alzhiemer’s Disease Progression

Stage	Description
0	No clinical or biomarker evidence of disease
1	Asymptomatic • Biomarker evidence of amyloid pathology • May have evidence of secondary pathology (elevated CSF tau or p-tau; hippocampal atrophy) • No cognitive or functional impairment present
2	Early symptomatic • Meets biomarker criteria for Stage 1 • Biomarker evidence of secondary pathology • Subjective and objective cognitive impairment • No functional impairment
3	Symptomatic • Meets criteria for Stage 2 • Definite cognitive impairment • Impairment in instrumental and some basic ADL
4.	Impaired • Biomarker evidence of amyloid pathology • Meets formal criteria for Dementia Syndrome
5	Incapacitating cognitive and functional disability

Enriching study populations for subjects likely to respond to a drug with a given mechanism of action theoretically should yield the greatest chance for success. Likewise, being able to stratify or otherwise correct for the contribution of mixed pathologies should increase the precision of endpoint analyses in trials. Finally, staging patients more accurately and precisely should increase the likelihood of executing a successful clinical trial. Several recent papers have explored the ADNI data set in attempts to supplement clinical criteria with biomarker criteria that might identify 1) subjects with the MCI syndrome likely to be the result of amyloid brain pathology (e.g. low A in the CSF (19) increased cortical amyoid burden demonstrated by PiB imaging (20) and 2) subjects at high risk for progressing to the stage of dementia within the near future (e.g. high CSF tau or p-tau, regional brain atrophy (21). We used such an approach to predict disease progression in the placebo group of a hypothetical clinical trial using the ADNI database*. By combining various clinical and biomarker criteria, we were able to distinguish MCI populations who progressed more rapidly on cognitive (ADAS-Cog), functional (FAQ) and global outcome measures (CDR-SB) than those who did not.

The current regulatory standards for AD clinical trial endpoints require demonstration of benefit on separate cognitive and functional tests (22). Analyses of ADNI data have demonstrated very clearly that the rate of progression of cognitive dysfunction in AD accelerates as subjects pass through the MCI syndrome into the stage of “mild” AD (20). The rate of change of the ADAS-Cog in MCI subjects is slow, and, given the variability inherent in the measure, is perhaps too slow to allow us to detect treatment effects in clinical trials of reasonable size and length. Other cognitive tests, such as the NTB ( 23., 24., 25.), have not been widely used in clinical trials in this population to date, and there is still some uncertainty as to the potential implications of practice and learning effects for a clinical trial when these tests are administered repeatedly over a relatively short period of time.

In symptomatic eAD, the emergence of functional deficits superimposed on impairment in two or more cognitive domains defines the dementia syndrome (2). Since we now view AD as a continuum of biological process that lead to cognitive and ultimately functional disability, the question arises, whether there might be a single assessment instrument that can capture the spectrum of disease progression in MCI on both cognitive and functional domains. For example all Parkinson’s Disease clinical trials utilize the Unified Parkinson Disease Rating Scale (UPDRS; 26). The UPDRS is a comprehensive assessment instrument that comprised of 4 parts, assessing behavior, motor function (assessed by examination), ADL function (assessed by patient/caregiver report) and complications of therapy. Most commonly, the sum of the scores on Parts II (motor) and III (ADL) of the UPDRS is used as a clinical trial endpoint. The UPDRS has been used as the primary endpoint for pivotal trials of putative disease-modifying (27) as well as symptomatic PD drugs.

The Clinical Dementia Rating Scale Sum of the Boxes (CDR-SB), is a commonly used and accepted comprehensive measure of cognitive and functional abilities in AD (28), and it has been demonstrated to be a useful staging instrument (29).

Given the strong face validity of CDR-SB, perhaps this instrument could be employed as a single outcome measure for pivotal trials in early or very mild AD patients, especially if supported by changes in one or more supported biomarkers (30). Trachtenberg et al (31) performed a factor analysis of the CDR-SB used in conjunction with other measures that separately rated cognitive function (MMSE), functional abilities (ADCS-ADL) and behavior (BRSD) in a sample of 242 subjects at 27 sites in the US with probable AD. A five-factor model accounted for 82.9% of the variance in the data. In this analysis, the authors created 2 CDR subscores: a “cognitive” subsum comprising the sum of the memory, judgment, and problem solving and orientation box scores of the CDR, and a “functional” subsum, which combined the scores for community activities, personal care, and home and hobbies boxes. In this model, the 12-month change in the CDR cognitive subsum loaded onto a factor with only the MMSE, and the change in functional subsum loaded onto a factor with only the change in ADCS-ADL score. The correlation between the change in the total CDR-SB and change in the MMSE was - 0.458, between CDR-SB and ADAS-Cog was 0.42 (p <.0001) and between the CDR-SoB total and ADCS-ADL the correlation was of a similar order, 0.502, indicating a “modest” degree of association between the change in the total CDR-SB score with decline in both standard cognitive and functional measures. Thus, the subdomain structure of the CDR-SB can be viewed as mapping onto the conceptual framework of Parts II and III of the UPDRS, namely assessment of signs (cognitive impairment for AD, motor signs for PD) and functional symptoms.

Use of the CDR-SB as a sole outcome measure for clinical trials in early or very mild AD is not without its limitations. It is a lengthy instrument to administer. Raters administering the instrument must be well-trained and adhere closely to the semi-structured interview format (32). And most importantly, the accuracy of the information obtained is highly dependent on the reliability of the study subjects’ study partner/caregiver/informant, and the consistency of a single informant being available throughout the duration of the clinical trial. However, inter-rater reliability studies have demonstrated high levels of intra- and inter-rater reliability in multicenter clinical trials (33).

Through the use of clinical staging and advances in biomarker and imaging technologies we may now be in a better position to select patients for early AD clinical trials than we were just a few years ago. We remain challenged to find sensitive cognitive and functional outcomes that are clinically meaningful at this stage of the disease. Initial analyses of the ADNI data suggest that the CDR-SB, may be well-suited for this purpose.