Brain Activation During Memory Tasks in People at Genetic Risk

      To determine the relationship between brain responses to memory tasks and genetic risk for AD, Bookheimer et al. (2000) performed apo E genotyping and functional MRI while cognitively intact older persons performed memory tasks. The study included 30 participants, aged 47 to 82 years, with mild memory complaints but normal memory performance, of whom 16 were apo E-e4 carriers and 14 were not. The age and prior educational achievement in the two groups were similar. Brain-activation patterns were determined from functional MRI scanning while participants memorized and recalled unrelated word pairs. Memory performance was reassessed on 14 participants 2 years later.

     The magnitude and spatial extent of brain activation during memory performance in regions affected by AD, including left hippocarnpal, parietal, and prefrontal regions, was greater in the participants with apo E-e4 alleles, as compared with those with no apo E-e4 alleles.

     During memory performance tasks, the apo E-e4 carriers demonstrated a greater percentage increase in hippocarnpal MRIsignal intensity and a greater number of activated regions throughout the brain than did participants without apo E-e4. Longitudinal assessment after 2 years indicated that greater baseline brain activation correlated with verbal-memory decline. These results indicate that brain-activation patterns during memory tasks differ according to genetic risk for AD and may provide information that eventually predicts future cognitive decline.

Cerebral Metabolic and Cognitive Decline in Apo E-e4 Carriers

     To determine cognitive and metabolic decline patterns according to genetic risk, Small et al. (2000) investigated cerebral metabolic rates using PET in middle-aged and older nondemented persons with normal memory performance.
      Participants were right-handed and in the 50- to 84-year-age range. Of the 54 participants with mild memory complaints, 27 were apo E-e4 carriers and 27 were noncarriers. A single copy of the apo E-e4 allele was associated with lowered inferior parietal, lateral temporal, and posterior cingulate metabolism, which predicted cognitive decline after 2 years of longitudinal follow-up. For the 20 nondemented participants followed longitudinally, memory-performance scores did not decline significantly,  but cortical metabolic rates did. In apo E-e4 carriers, a 4% left posterior cingulate metabolic decline was observed, and inferior parietal and lateral temporal regions demonstrated the greatest magnitude (5%) of metabolic decline after 2 years. These results have practical implications for clinical trials of dementiaprevention treatments. The right lateral temporal metabolism for apo E-e4 carriers at baseline and 2-year follow-up yielded an estimated power under the most conservative scenario (i.e., assuming that the points are connected exactly in reverse order) of 0.9, to detect a 1-unit decline from baseline to follow-up using a one-tailed test. A sample size of only 20 participants, therefore, would be needed in each treatment arm (i.e., active drug or placebo) to detect a drug-effect size of 0.8 (a = 0.05, power = 0.8).

     Thus, a clinical trial of a novel intervention to prevent cerebral metabolic decline would require only 40 participants over a 2-year treatment period. Such findings are consistent with previous PET studies showing stable and replicable results (Andreasen, Arndt, Cizadlo, et al., 1996), and suggest that combining PET and AD genetic-risk measures will allow investigators to use relatively small sample sizes when testing antidementia treatments in preclinical AD stages.

     These results indicate that the combination of cerebral metabolic rates and genetic-risk factors provides a means for preclinical AD detection that will assist in response monitoring during experimental treatments.

Brain Imaging Surrogate Markers for Detection and Prevention of Cognitive Aging and Alzheimer's Disease

       With increasing knowledge of the neuropathological and cognitive changes preceding Alzheimer's disease (AD), clinical trials have begun to focus on preventive treatments aimed at slowing age-related cognitive decline and delaying onset of AD.

      Studying participants with minimal deficits leads to diagnostic heterogeneity and a need for larger samples in order to detect active drug effects. This chapter summarizes results of several studies designed to address such issues. Middle-aged and older adults with mild memory complaints were studied using brain imaging and measures of the major known genetic risk for AD, the apolipoprotein E-4 (apo E-e4) allele.

      In a study of positron emission tomography during mental rest, glucose metabolic rates were significantly lower in apo E-e4 carriers in brain regions affected by AD. Another study using functional magnetic resonance imaging showed increased brain activation during memory tasks in apo E-e4 carriers in similar brain regions. Longitudinal follow-up after 2 years indicated the potential utility of such brain-imaging measures, combined with genetic-risk information, as surrogate markers in prevention-treatment trials for cognitive aging and AD.

      Additional development of novel approaches using positron emission tomography to directly measure the neuritic plaques and neurofibrillary tangles of AD offers promise of more specific measures of disease progression in future clinical trials.

The fourth method

     The fourth method produced population-based, average 3-D hippocampal surface maps for both the control and MCI groups, derived from the manual outlines produced across centers. The methodology for constructing a surface map was described and was initially applied to document the variability of the elderly control, MCI, and AD hippocampus in the Talairach atlas
space, arguing for a refinement of this commonly used space in the functional imaging assessment of the elderly and demented population. Hippocampal
contours were traced using a track ball on each relevant slice. The points that made up a traced contour, after smoothing the effects of irregular hand movements, were connected across slices to create regularly ordered 3-D meshes corresponding to the hippocampal surface. To quantify an individual patient's anatomy an average hippocampal 3-D model was resolved by averaging the vectorial displacements, on a point-by-point basis from the parametric mesh, for each patient's hippocampus.

    Hippocampal variability, expressed as a 3-D distance in the common coordinate space, was computed by taking the mean of the square roots of a 3-D displacement vector necessary to align each node along a patient's mesh onto the average representation of the population's hippocampus.

A linear measure of medial temporal lobe

   The third method was a linear measure of medial temporal lobe width that used the centimeter-distance scale of images that have the midpoint of the rostral caudal extent of the hippocampal formation at each side.

    This linear measure of the narrowest point of the medial temporal lobes adjacent to the brain stem used the procedure described by Jobst et al. (1998,1992). Image was accessed via the Internet, and measures were entered electronically over the web.

Assessment methods.

       Four assessment techniques were implemented at each center. The first was a traditional volumetric assessment technique with manual outline of left and right hippocampi according to the methods of Jack et al. (1990,1989). Contours were made using a Java-based outlining package distributed to all centers over the Internet, which can be compiled on all current personal computers, laptop computers, or unixbased workstations. The digital output of these contours were electronically sent to UCLA for processing.

    The second procedure was to reformat the scans axially, parallel to the long axis of the hippocampus, according to the methods of de Leon et al. (1993) and to provide an individual slice through each volume to obtain qualitative assessments of medial temporal atrophy using a 4-point scale. The images of each slice were accessed via the Internet, and scores were entered electronically over the web.

Statistical analysis.

     Volumetric assessment included normalization for interpatient variation of head size by dividing hippocampal volume by the total intracranial volume of that particular patient. Percentiles for normalized hippocampal volume had been previously reported. Percentile scores were then converted to a W score, which is a value from a standard normal distribution corresponding to the observed percentile (in a standard normal distribution, the 50th, 5th, and 2.5th percentiles are given by W scores of 0, -1.645, and -1.96, respectively).

       A W score of less than or equal to -2.5 would be associated with the 0.6th percentile of hippocampal volume. Sixty-seven of 80 MCI patients in a recent study (Jack et al., 1999) had W scores less than 0 (13 had W scores less than or equal to -2.5, which is 16.25% of the population of MCI patients studied). Assuming outliner variability as previously reported across studies of errors less than 1.9% coefficient of variation, then the number of MCI patients needed to produce sufficient power to find a significant effect would be 30.
       This would provide 80% power to detect a difference of 16% as a deviation from .5, with alpha equal to .01. If 10.5% of the 30 MCI patients fall below a W score of -2.5, rather than the expected 16.25%, we would have 80% power to detect an alpha at .01 (h = .45 at 0.6th percentile of the control's standard normal distribution) (Cohen, 1988). Sensitivity (true positives / (true positives + false negatives) and specificity (true negatives/ (false positives + true negatives) calculations were also computed for each method's ability to correctly identify incipient AD, MCI, and controls.

Techniques for identification of patients with MCI compared with controls

    The four techniques for evaluation were chosen based upon their demonstrated usefulness in evaluating medial temporal atrophy. These include:
1. The volumetric assessment technique of the Mayo group (Jack et al.,1990,1992).
2. The qualitative assessment of medial temporal lobe atrophy performed by the NYU group (de Leon et al., 1993,1989,1997).
3. The linear assessment technique of Jobst et al. (1998,1992).
4. The three-dimensional (3-D) mapping technique employed at UCLA (Thomspon & Toga, 1996; Thompson et al., 1996). The centers participating in this multisite protocol who share both common MCI definitions and MRI acquisition sequences are listed in Table6.1, along with the number of MCIs and controls to be studied.

Mild cognitive impairment (MCI) may have incipient Alzheimer's disease

    Elderly individuals with mild cognitive impairment (MCI) may have incipient Alzheimer's disease (AD) and are the most likely targets for therapies directed at disease modification or slowing progression to the
endpoint of overt dementia. Identification of those individuals who will eventually develop AD is assuming urgent importance in many centers.

    With the potential advent of disease-modifying drugs, biomarkers with good reliability, sensitivity, and specificity are needed. One promising biomarker for patients with incipient AD is hippocampal atrophy
assessed by magnetic resonance imaging (MRI).

Magnetic resonance

      Magnetic resonance imaging has been used routinely in the past to ensure diagnostic accuracy by ruling out stroke and other structural causes of cognitive impairment. Present technology, however, allows volumetric measurements of brain structures to be performed during the course of a trial that may serve additional purposes. Volumetric measurements obtained at the inception of a trial provide indices of whole brain and regional atrophy that relate to the fundamental pathology of the disease process. They offer an alternative approach to cognitive and clinical assessments for staging patients. They may be helpful in predicting disease progression and serve as potentially important explanatory covariates in efficacy analyses of therapeutic agents.

    Serial MRI volumetric measurements performed over the course of a trial permit investigators to directly monitor volume loss resulting from ongoing pathology, and may help distinguish whether a therapeutic agent has symptomatic or diseasemodifying effects.

     The relationships found so far between the hippocampal volume and the clinical assessments obtained in participants upon entering the MCI trial further point to the utility of MRI volumetric analysis as a valuable indicator of disease severity. They suggest that even within the relatively narrow continuum of MCI participants, MRI volumetric analysis can discriminate between participants on the basis of their performance on neuropsychological and clinical evaluations. These encouraging results support the contention that baseline MRI volumetric analysis may predict subsequent conversion to AD and be useful for measuring the effects of treatment.

Volumetric analyses have been performed on MRI

     Volumetric analyses have been performed on MRI scans obtained at the time of study entry from the first 77 participants enrolled in the MRI portion of the MCI protocol. Fifty-eight percent of these participants are men and 42% are women. They have a mean age of 73 years and a mean education of 15 years. They have a mean score of 28 on the Mini-Mental Status Examination (MMSE) (Folstein, Folstein, & McHugh, 1975), a mean score of 11 on the ADAS-COG, and a mean score of 1.8 on the CDR Sum of Boxes.

     We performed correlational analyses between the MRI hippocampal volumes at study entry and the participants' concurrent performance on (ADAS-COG) and functional (CDR Sum of Boxes) measures. The results indicate that hippocampal volume in these participants predicted performance on the ADAS-COG (r = -.31, P < .01) and the CDR Sum of Boxes (r = -.33, P < .01). These correlations remained significant after controlling for age and gender. The relationships are shown graphically in Figures 5.1 and 5.2. For illustration purposes, the MRI hippocampal volumes are divided into tertiles from smallest to highest.

intrarater test-retest coefficient of variation of hippocarnpal volumetric measurements is 1.9%

     Validation studies demonstrate that the intrarater test-retest coefficient of variation of hippocarnpal volumetric measurements is 1.9% with this method (Jack, Bentley, Twomey, & Zinsmeister, 1990). The 3D-image data set of each patient is realigned into an orientation perpendicular to the principal axis of the left hippocarnpal formation. The imaging data are then interpolated in-plane to the equivalent of a 512 x 512 matrix and magnified two times. The voxel size of the fully processed image data is 0.316 mm3. The borders of the hippocampi are manually traced on the workstation screen for each image slice sequentially from posterior to anterior.
     Typically, 40 to 50 imaging slices are measured for each hippocampus. In-plane hippocampal anatomic boundaries are defined to include the CA1 to CA4 sectors of the hippocampus proper, the dentate gyrus, and the subiculum (Jack et al., 1997). The posterior boundary of the hippocampus is determined by the oblique coronal anatomic section, in which the crura of the fornices are defined in full profile.

     Pearson correlation coefficients were calculated between the hippocampal volume measured from MRIs obtained at study entry, as well as the participants' baseline Alzheimer's Disease Assessment Scale-Cognitive portion (ADAS-COG) (Rosen, Mohs, & Davis, 1984) and Clinical Dementia Rating (CDR) Sum of Boxes (Morris, 1997), in order to determine the relationship between hippocampal volume and cognitive and functional performance at baseline.

treatment with vitamin E, donepezil, or placebo

         Approximately 720 participants with MCI are currently being randomized to treatment with vitamin E, donepezil, or placebo for 3 years, with roughly an equal number of participants (240) in each treatment group.
      From previous data, it is known that individuals who meet criteria for MCI convert to AD at a rate of approximately 12% to 15% per year (Petersen, Smith, Ivnik, et al, 1995; Grundman, Petersen, Morris, et al., 1996b; Petersen et al., 1999). Additional data suggest that this conversion rate may be even higher (up to approximately 20% per year over the first 3 years) in MCI participants who are screened with a memory test assessing delayed recall. Based on these figures, it is expected that at the end of the 3 years, 45% to 60% of placebo-treated participants may develop AD.

     The trial is designed to evaluate whether the two treatment interventions are effective at reducing the rate of conversion to AD by approximately one third over the course of the 3 years. The MRI component of the protocol stipulates that MRI scans are obtained on participants at entrance into the trial, upon completion of the trial 3 years later, and at the time of diagnostic crossover to AD. In all participants the MRI study is performed according to a defined set of imaging sequences that are sufficiently generic so that they can be executed across a variety of MRI vendors. The cornerstone of the planned volumetric analyses includes a 3-dimensional (3D) volumetric imaging sequence that provides optimal spatial resolution in all three anatomic dimensions.

     Guidelines for the storage and archival procedures are provided to each of the individual participating sites. Magnetic resonance imaging data cassettes from participating sites are sent to the central data repository at the Mayo Clinic in Rochester, Minnesota, where imaging data are downloaded, checked for compliance with the prescribed imaging sequences, and then catalogued. Catalogued images are cross-checked with case report forms forwarded from participating clinical sites to the ADCS coordinating
center in San Diego, California. After the images are analyzed, the data are merged with other clinical data on each participant residing in the ADCS central database.

    The proposed analyses focus primarily on the medial temporal lobe, particularly the hippocampal formation. Nevertheless, with the imaging sequences outlined in the study protocol, the capability exists to measure serial whole brain, lobar, sublobar, and ventricular volumes, as well as gray and white matter volumes, and leukoaraiosis. Based on prior studies, we anticipate that the hippocampal volume or other baseline measurements on the baseline MRI scans will predict which participants are most likely to develop AD. Analyses of the serial images obtained over the course of the study will examine whether treatment with either vitamin E or donepezil alters the rate of atrophy in the whole brain, lobar and sublobar-medial temporal lobe, entorhinal cortex, as well as the gray and white matter. Similar analyses will examine whether there are treatmen differences in the rate of ventricular enlargement.

The earliest pathological changes in the brains of AD

    The earliest pathological changes in the brains of AD patients occur in the medial temporal lobe. The entorhinal cortex and its perforant pathway projections to the hippocampus are among the earliest brain regions to exhibit the neuropathological hallmarks of AD (Braak & Braak, 1996). The resulting loss of neurons and synapses in this region is manifested at a macroscopic level by medial temporal lobe atrophy, which can be observed by visual inspection of the brain at autopsy or by MRI in living patients.

    Paralleling the neuropathological changes, clinical symptoms of AD often involve cognitive impairments and fluctuations in body weight that are closely linked to medial temporal lobe atrophy (Grundman, Corey-Bloom, Jernigan, Archibald, & Thai, 1996a). In recent years, considerable effort has been devoted to imaging these structures to see if they serve as a useful adjunct in diagnosing and following patients with early AD.
    Previous work demonstrates that MRI-based quantitative measurements of the hippocampal formation are useful in differentiating patients with AD from normal control participants (Jack, Petersen, Xu, et al., 1997).
They found that the best discriminator between AD patients and normal controls was the hippocampal volume when different medial temporal lobe structures were compared. Participants with mild AD had mean hippocampal volumes that were approximately two standard deviations below the control mean.

   Volumetric measurements of the hippocampal formation may also be useful for predicting future conversion to AD in MCI participants. Jack, Petersen, Xu, et al. (1999) measured MRI-based hippocampal volume in
80 consecutive patients who met criteria for the diagnosis of MCI, and followed them longitudinally with approximately annual clinical and cognitive assessments. The primary endpoint was the crossover of individual MCI patients to a clinical diagnosis of AD during follow-up.

    During the period of observation, which averaged 33 months, 27 of 80 MCI patients developed dementia. Hippocampal atrophy at baseline was significantly associated with subsequent conversion from MCI to AD. The risk of AD was 50% within 3 years in MCI patients with moderate hippocampal atrophy; however, it was only 26% in patients with mild hippocampal atrophy (approximately 50% reduction in risk of conversion based solely on differences
in hippocampal volume). In MCI patients with normal hippocampal volume, the risk of developing AD dropped even further, with only 9% crossing over to AD within 3 years.

Orientation-Memory-Concentration test

    Approximately 830 individuals over age 75 years were on the computer roster of a large outpatient clinic affiliated with a managed care organization in Minneapolis, Minnesota. We selected those individuals from this roster without diagnoses of dementia, Alzheimer's disease (AD), and their synonymous diagnostic codes in any patient encounter in the prior 2 years. Of the 611 individuals who met these criteria, we sent an introductory letter to 447 potential participants, informing them that they would be receiving a phone call as part of this study.
    Upon reaching a participant on the telephone, the examiner (an RN experienced in geriatrics and formally trained to administer the mental status examinations) informed the individual of who she was and the
nature of the study. For those who agreed to participate, they were given either the Orientation-Memory-Concentration (OMC) test (Katzman, Brown, Fuld, et al., 1983) or the Minnesota Cognitive Acuity Screen (MCAS) (Knopman, Knudson, Yoes, & Weiss, 2000). Administration of the two generally alternated, except in a few circumstances in which a participant would agree only to the shorter of the two tests.

     After completion of the mental status examination, the participants were questioned on four issues. The first question asked how they thought they performed on the mental status examination ("How do you
think that you did on the tests we just finished?"), and the responses were good, fair, poor, or very poor. The second question was, "Apart from this interview, in the past few months, have you felt that you have had memory problems?", with the rating scheme as follows: "my memory is excellent," "my memory is as good as anyone else my age," "my memory isn't as good as it used to be, but it doesn't cause me any problems," "my memory has definitely deteriorated in the past few years and interferes with my daily activities," or "my memory is severely impaired." The third question was whether they would follow-up with their physician if the  were told that they had done poorly ("Suppose that the present interview showed that you had memory problems, would you be willing to have a medical assessment for memory problems at your clinic?", with yes, no, or possibly as the responses.
      The final question was open-ended, asking what they thought of the process of telephonic examinations. The study was approved by the HealthPartners Institutional Review Board. Oral consent was obtained to conduct the phone interview.

Disscusion

       Under conditions that might be present in many clinical practices, the acceptance/completion rate for telephonic mental status assessment among participants 75 years and older was only 55%. Of those who underwent cognitive assessment, the yield of cognitively impaired participants was low. Only 3% scored in the range that was indicative of cognitive impairment.

      Among the interviewees, willingness to consult with a physician regarding memory problems detected during a telephonic screen was quite high (87%).

       The low number of individuals who scored in the impaired range suggests that the telephonic strategy did not serve to uncover undiagnosed cognitive impairment in this elderly cohort. We believe we should have
encountered 20 screen failures (roughly a 10% prevalence in this 75+ year only age group), in contrast to the seven we found.
       We suspect that the lower-than-expected number of cases occurred as a result of selective refusal by participants with cognitive difficulties to participate in the telephonic screening. Unfortunately, we have no way of determining the cognitive status of those individuals who refused to be interviewed. From other studies in which cognitive status was available from an earlier visit prior to a refusal to continue to participate, those who refused to participate in follow-up had lower performance at baseline. In the Canadian experience, of 147 participants whose cognitive screening was abnormal, but who refused to undergo a clinical examination, their mental status score was more than seven points lower on the Modified Mini-Mental State examination, compared with those who continued to participate in the study (81.3 vs. 88.8).
     The low overall rate of completion of interviews could be attributed to a basic flaw in the telephonic strategy. In current American culture, telephone marketing is widespread, and many people, upon receiving phone
calls from strangers, may resent the intrusion. An alternative strategy would have been for us to have engaged the primary physicians in this clinic and encouraged them to promote participation. Because patients might see their physicians only every 6 months or less, such an approach would work only if our project had had a longer time scale.
      The findings of the present study suggest to us that telephonic screening is a useful and efficient strategy to identify those who do not need face-to-face cognitive assessments. Assuming that telephonic screening
were to be identified to the health plan membership as "required" or "strongly recommended," as opposed to an optional activity tied to research as was necessarily the case with the present project, the completion rate
could be higher. The number of elders who would then need face-to-face examinations would be perhaps only a half or a third of all individuals over age 75 years.

Magnetic resonance imaging (MRI)

       Magnetic resonance imaging (MRI) volumetric analysis is a relatively new assessment tool in clinical trials for Alzheimer's disease (AD). It is a technique that is particularly well suited for clinical trials designed to prevent AD or slow its progression. The Alzheimer's Disease Cooperative Study, a multicenter consortium that performs AD-related clinical trials, is presently conducting a clinical trial with vitamin E and donepezil (Aricept®) in participants with mild cognitive impairment (MCI). The objective of the trial is to determine if either agent is capable of delaying a diagnosis of AD.

          Mild cognitive impairment participants are randomized to treatment with vitamin E, donepezil, or placebo, and followed longitudinally for 3 years. A subset of randomized participants is undergoing MRI of the brain prior to beginning treatment and during the course of the trial. The objectives of the MRI component of this trial are to determine (a) whether MRI volumes predict cognitive and functional performance in a well-defined cohort of participants with MCI; (b) whether hippocampal volume or other baseline volumetric measurements predict subsequent crossover to clinical AD; and (c) whether treatment with either vitamin E or donepezil alters the rate of brain atrophy over the course of treatment.

         Although the trial is still recruiting participants, analysis of the MRI scans collected thus far indicate that hippocampal volumes measured at the time of study entry predict baseline cognitive and functional performance. These results support the hypothesis that brain volumetric analysis may also be useful to predict future prognosis and for monitoring the effects of treatment on disease progression.

Autosomal recessive genes

    Consanguinity increases the prevalence in a population of autosomal recessive genes that may have undesirable characteristics. For example, an additional 1/16 of the variation of DNA is made homozygous by the inbreeding of a marriage of first-cousins. Conversely, the probability of identifying recessive or quasirecessive susceptibility factors is enhanced by inbreeding. Childhood mortality is increased in offspring of first-cousin marriages by a factor of 1.4 to 1.7, and children born to consanguineous unions have poorer health than the offspring of nonconsanguineous unions (including malignancies, congenital abnormalities, mental retardation, and physical handicap). Curiously, there are few studies of the effects of inbreeding on the health of adults. We have hypothesized that consanguinity in the Wadi Ara community has increased the prevalence of autosomal recessive genes that are responsible, in part, for the increased prevalence of the disease.
    It has been reported that 44% of all Arab marriages in Israel are consanguineous, with a mean inbreeding coefficient of .0192 (Jaber, Shohat, Rotter, & Shohat, 1997). The inbreeding rates for Israeli Arabs may be particularly high (as compared with Egyptians or Syrians) because mobility was reduced for centuries by the Turks. At times in Arab communities, it has been required for children to marry into their own family in part to avoid sharing resources with competing family groups.

All of the known Alzheimer-related genes (on chromosomes 21, 14, 1, and 19) are dominant (chromosomes 21,14,1) or codominant (chromosome 19). There are no known genes affecting the development of AD that are recessive, perhaps because there have been few studies of AD in populations with high levels of inbreeding. De Braekeleer et al. (1989) reported an association between inbreeding and the development of AD (inbreeding coefficient 9 times higher in AD cases than in controls) in a rural population in Quebec. Studies in the Old Order Amish have found a low prevalence of disease in this inbred community, also having a low frequency of the apo E-e4 allele (.037). Inbreeding could be linked to AD through a confounding effect of education, but the association of inbreeding and education is controversial (found for a Saudi population but not for Israeli Arabs). We expect that there are recessive factors for AD because of indirect evidence: genetic modeling studies of AD in apo E-e4 negative families are unable to reject a recessive model (Rao et al., 1996).
    However, it is difficult to evaluate recessive models in outbred populations. The high prevalence we have seen in our highly inbred Arab population in Israel might be due to the inculcation of recessive AD susceptibility alleles or to a dominant gene that became frequent by founder effect.