Working Group on Study Design - Summary of Findings

NIA AD Neuroimaging Initiative

October, 2002

Working group members participating:                   

Peter J. Snyder (Chair)
Leon Thal        
Marilyn Albert
Orest Hurko
Mark Schmidt
Ronald Petersen
John Morris
Richard Hargreaves
Mike Grundman

1.  Goals and Objectives of NIA Neuroimaging Intiative:

1) To both identify such novel neuroimaging markers of disease progression, and to arrive at accurate estimates of the variance of, and confidence intervals for, their measurement.

2) To explore the potential additive value of multiple and complementary imaging modalities (e.g., sMRI and FDG-PET) in developing biomarkers that may rely on such multi-spectral analyses.

3) To identify the relationship between change in these key neuroimaging markers with clinical change, as well as to corresponding change in key biochemical measurements, over the course of disease progression (over a 3 year study period).

4) To understand what portion of the variance in change over time for several key neuroimaging markers (e.g., hippocampal volume on MRI, or biparietal hypometabolism as measured by FDG-PET) is accounted for by individual differences in known genetic risk factors.

5) To develop estimates of the utility of measures of brain structure and function, genotype, and biochemical markers for the prediction of the onset and progression of Alzheimer's disease.  The goal of this work shall be to identify and begin validating suitable surrogate markers of AD progression.  A further goal is to identify markers that sensitively predict the risk of conversion from MCI to AD.

6) To put into practice, and to gain experience with, the necessary imaging QA/QC procedures that are critical to adhere to in this field, within the context of large multi-center trials.  To agree on common methods for the careful reduction of instrumentation, calibration, and measurement error throughout the course of a longitudinal study.

7) To create a “gold-standard” library of imaging, clinical, neuropsychological, and body fluids assays from three key subject groups, with the principal aim of targeting the transitions from healthy control (HC) to MCI, and from MCI to AD.  This library will then serve as a “test bed” for a wide variety of imaging algorithms and methods as they are created.  A common library will also serve to explore the inter-rater reliability for such measurements (both automated and non-automated) across centers and investigators.  This library would be the core “product” created by the Consortium, and would be publicly available for exploration and data mining.

2.  General Study Management Issues:

A. Central Site Management –

• A single central coordinating center will be identified to manage the transmission & storage of clinical, biological fluids, genotyping, and imaging data. 
• The central coordinating center may not be responsible for all aspects of QA/QC.  For instance, the site that is best prepared to manage continuous QA/QC for MRI imaging will likely not be the same site that is best qualified to manage these same issues for PET imaging.  However, there will be one central coordinating center that is responsible for managing where these discrete portions of the larger database are located, and will be responsible for ensuring that all portions of the database will be brought together and archived on a regular basis (to be determined).
•  The central coordinating center will be responsible for tracking subject enrollment into all groups, and for ensuring that group-to-group matching (age, sex, education, race/ethnicity) occurs.

3.  Determination of Subject Groups:

Specific Inclusion and Exclusion Criteria for all three Subject Groups to be Provided in Separate Excel Spreadsheet.

Several of the key points are as follows:

A. Inclusion of a Mild AD Group –

• The decision was made to include a 3 rd group of mild AD subjects in study, principally to map the rate of change of specific brain ROIs, in this stage of the illness, that may reflect neuropathologic changes that are specific to this phase of the illness.  Mild AD subjects are still intensively studied in clinical trials, and the rate of CNS changes may be accelerated in this stage (relative to MCI).
• Mild AD group shall be defined, in part, as having MMSE scores on entry of between 18 and 26, inclusive.

B. Inclusion/Exclusion Criteria –

• General agreement for modeling MCI inclusion/exclusion criteria after those employed by the ADCS for the ongoing “Vitamin E” study.  Some of the medication exclusions may need to be deleted, although there is general agreement that the use of either neuroleptics and/or antiepileptic drugs (AEDs) should remain as exclusion criteria.  The ADCS study has found, with these criteria, a conversion rate of 10-14% per year, and a cumulative conversion rate (from MCI to AD) of about 35-40%.
• Inclusion & Exclusion Criteria Are Provided in Separate Spreadsheet

C. Matching of Groups –

• The HC and MCI groups will be matched, at the group level but not on a one-to-one basis, by age, sex and education.  The central coordinating center will monitor enrollment and alert sites when a particular “cell” has been filled (e.g., when no more males over the age of 70 with 14+ years of education should be enrolled into the HC group).
• No attempt will be made to match by SES per se , although matching by education will largely serve a similar function.
• ApoE genotyping should be completed at the screening or baseline exam.  Subjects should not be excluded or included on the basis of genotype, but an attempt will be made by the central coordinating site to match the ratio of ApoE e4 positives to negatives, so that ideally the HC group will have at least 15% ApoE e4+ subjects.

D. Inclusion of Minority Groups –

• For all three subjects groups, a concerted effort will be made to ensure that 1 out of every 10 subjects enrolled belongs to a minority (race/ethnicity).  The central coordinating center will monitor enrollment and alert sites when a particular “cell” has been filled.
• All cognitive assessments may be completed in either English or Spanish (when necessary), and all individual tests that are recommended for this study have been previously been translated and validated for use in Spanish – with the one exception of the Paired Associates List (PAL) test.  For the PAL test only, this will be considered to be an “exploratory endpoint” for those cases in which it was administered in Spanish.
• Subjects with Major Depression will be allowed in the study, as long as they are being successfully treated for this condition.
• Mention of laboratory studies will be moved from inclusion to exclusion criteria.
• Include MRI exclusions (cochlear implants, pacemakers, clips in head…)
• Exluded: History of schizophrenia or other chronic psychotic disorder (eg, schizoaffective, delusional disorder); current untreated major depressive episode.

4. Broad Study Design:

A. Duration of Longitudinal Follow-Up –

• Three (3) years for MCI and HC groups
• Two (2) years for the AD group

B. Timing of Visits –

This committee is recommending study visits at:

HEALTHY CONTROLS: Baseline, 12 month, 24 month, & 36 Month Visits
MCI GROUP: Baseline, 6 Month, 12 Month, 18 month, 24 month, 30 month, & 36 months Visits
MILD AD GROUP: Baseline, 6 Month, 12 Month, 18 Month, & 24 Month Visits 

In the best case, at each time point all subjects will complete clinical, body fluids collection, neuropsych. and imaging evaluations.  In practice, this will not be possible with all endpoints.  An effort should be made to come as close as possible to this goal.

In general, more frequent contacts with subjects is desirable, and will lead to better subject retention in the study.

C. Subject Compensation and Travel Assistance –

• Committee agrees that it will likely be necessary to provide some level of compensation to subjects for their participation.
• It is also likely that travel assistance for subjects to visit medical centers for study visits may be necessary on a case-by-case basis (e.g., taxi services).
• Both of these items need to be budgeted for in the final study budget.

D. Screening Exam(s) –

• Screening visit(s) will be important for several reasons: 1) to complete a brief neuropsych. assessment to identify subjects who meet entry criteria for the MCI group; 2) to screen for neurologic and/or psychiatric diseases that are exclusion criteria; and to administer all neuropsych. measures in the study battery at least twice prior to the true baseline exam.  Both of these “practice” administrations may be completed on the same day (e.g., morning and afternoon of the screening visit), but ideally these should be completed on separate days.  These practice assessments must be done in order to minimize the impact of ‘task familiarity effects' that would confound the neuropsych. results obtained at baseline and on subsequent visits.

E. Clinical Treatment During Study Period –

• Committee agrees that this will be a naturalistic progression study.  No attempt will be made to control medications or other treatments that any subjects receive during the study.  However, all treatments administered to all subjects must be tracked (including vitamin and herbal therapies), so that these variables might be used as potential covariates during any data analyses.
• It is important to track all drug use, throughout study, for all subjects – for potential use as covariates in any post hoc statistical analyses.

5. Sample Size Determination:

Using the numbers below, assuming 15% drop rate per year, we should have about

215 MCI at the end of 3 yrs. Total number of converters would be about 45%

of 350 = 157.  All subjects who terminate early should have a termination scan & study visit at whatever time they withdraw from the study. 

Group  Duration(yrs)    N at start          N at end

Controls           3          150                  92

AD                   2          150                  108

MCI                 3          350                  215

6. Neuropsychological Batteries:

A. The decision was made to not rely on use of alternate forms for any of the neuropsychological tests.

B. The word-list learning test should be use an unstructured list (e.g., without obvious semantic categories), such as the Rey Auditory Verbal Learning Test (RAVLT). 

C. There should be one practice session with these tests, at screening (see below), prior to the “true baseline,” in order to reduce influence of task familiarity and practice effects.

D. The Geriatric Depression Scale (GDS) and State-Trait Anxiety Index (STAI)  will not be used.  Instead, the Hamilton Depression Scale and the Neuropsychiatric Inventory (NPI) will be completed at baseline and at 12, 24, and 36 month visits – these will provide ongoing estimates of depression and anxiety.

E. A subjective rating scale of sleepiness will not be necessary.

F. At the screening exam, subjects in all three groups will complete the MMSE, the CDR, the WMS Logical Memory subtest (required to meet inclusion criterion for MCI), the PAL, and a word-list learning Test (e.g., RAVLT).

G. At baseline exam only, subjects will complete the A-NART (considered to be a “hold test” that provides a rough approximate estimate of premorbid IQ)

H. For all other exams, starting with the baseline exam, subjects in all three groups will complete the:

• The CDR, and the Sum of Box Scores
•  ADAS-Cog
•  MMSE
•  The MCI-ADL Scale (completed with caregiver)
• The Neuropsychiatric Inventory (only at baseline, 12, 24, & 36 months)
• Hamilton Depression Rating Scale (21 item version?) (only at baseline, 12, 24, & 36 months)
•  Trailmaking Test, Parts A & B
•  The Rey Auditory Verbal Learning Test
• The WMS-III Paired Associates Learning Test

7.  Additonal Notes/Comments/Requests:

A. This committee recommends that the NIA convene a 3 rd Advisory Board Meeting, to take place several weeks after an initial protocol or RFA has been drafted.  This will allow all members of all 4 working groups to meet together to agree on any diverging recommendations between committees and to tie-up any loose ends, as this is will be a very costly, complex, and labor-intensive undertaking.

B.  There is general agreement that there should be a built-in phase that allows for the development and piloting of the imaging methods (MRI and PET), data transmission methods, QA/QC methods, and the piloting of key neuropsychological instruments.  The duration and management of this “pilot measures phase” remains to be determined.  An alternative to this approach would be to initiate a separate pilot study to resolve several methodologic issues (Item C below, for example) prior to the start of the larger trial.

C.  At the very least, a substantial proportion of centers involved in this study should have 3T field-strength magnets. By the time that this study concludes, 3T magnets will likely be the new standard for clinical MRI imaging, and this should be anticipated in this study.  Currently, there are approximately 100 centers worldwide that have 3T magnets, and the major MRI equipment manufacturers predict that the number of such centers will grow dramatically over the next 5 years.  The study design should incorporate a method for “bridging” from 1.5T to 3T.  Although is debate as to whether superior brain images might result from imaging with a 1.5T magnet instead of a 3T magnet, such problems with 3T may result from poor coil design and/or inadequate correction for field inhomogeneities. The major equipment manufacturers (e.g., GE, Siemens, Philips), as partners in this consortium, may be instrumental in helping to resolve these issues. The NIAMS multicenter Osteoarthritis Initiative has apparently decided to use 3T magnets exclusively, and GE has offered participating centers a preferred (discounted) rate to upgrade their magnets.  Perhaps we might want to evaluate the possibility of a similar arrangement (with GE, Siemens, and/or Philips) for the NIA Imaging Consortium as well.

D.  A variety of reasonable imaging endpoints have been proposed (cf. hippocampal volumes, boundary shift measurements), but it will be important to obtain these measures from more than one independent laboratory, and using more than one set of algorithms for volumetric analyses. One of the key goals of this study should be to explore the variance in measurement of key brain structures by independent investigators that are using a common database.  Hence, all imaging data need to be collected in a manner that would allow for unbiased reconstruction, so as not to favor any one method or hardware package.