|Home | About | Journals | Submit | Contact Us | Français|
First-degree family history of sporadic Alzheimer Disease (AD) and the apolipoprotein E ε4 (APOE4) are risk factors for developing AD. Although the role of APOE4 in AD pathogenesis has been well studied, family history remains a rarely studied and poorly understood risk factor. Both putatively cause early brain changes prior to symptomatic disease, but the relative contribution of each on brain function is unknown. We examined 68 middle-aged participants with a parent diagnosed with AD (+FH) and 64 age- and education-matched controls without a first-degree family history of any dementia (−FH). All underwent cognitive testing, APOE genotyping and a functional MRI (fMRI) encoding task that required discrimination of novel items from previously learned items. A 2 × 2 factorial ANOVA (presence/absence of parental family history and presence/absence of the APOE4) was used to detect group effects. A greater response to novel items was detected in the mesial temporal lobe and fusiform gyrus bilaterally among persons without a first-degree family history of AD. In hippocampal areas, the −FH +ε4 group exhibited the greatest signal change, and the +FH +ε4 group the least. These findings indicate that FH of AD is an important predictor of hippocampal activation during encoding and that FH may modulate the effect of APOE4 in these middle-aged adults, suggesting that an as yet unspecified factor embodied in first-degree family history of AD is influencing the expression of APOE4 on brain function.
There is increasing evidence that the clinical syndrome of Alzheimer disease (AD) is preceded by a silent preclinical phase characterized by neuropathological change spanning over decades (Ohm et al., 1995; Braak et al., 1999; Corder et al., 2004). Both the apolipoprotein E (APOE) epsilon4 allele (APOE4 or ε4) and first-degree family history (Fratiglioni et al., 1993) increase the pre-clinical prevalence of neurofibrillary tangles and amyloid plaques, the main pathology in AD (Ghebremedhin et al., 1998; Ghebremedhin et al., 2001; Corder et al., 2004). Additional evidence of pre-clinical AD comes from studies of cognitively normal APOE4 middle-aged adults who exhibit reduced cerebral metabolic rate of glucose (CMRgl) in many of the same regions affected by AD compared to non-carriers (Reiman et al., 1996; Small et al., 2000; Reiman et al., 2004). In older adults, APOE4 has been associated with decreased mesial temporal lobe (MTL) volume (Plassman et al., 1997; Lemaitre et al., 2005) and cognitive decline (La Rue et al., 1995; Caselli et al., 1999; Small et al., 2000; Caselli et al., 2001; Baxter et al., 2003; Blair et al., 2005). However, as a predictor of AD (Martinez et al., 1998; Green et al., 2002; Cupples et al., 2004; Pedersen et al., 2004), APOE4 accounts for less than 40 percent of AD cases, suggesting that other factors are involved in AD pathogenesis (Saunders et al., 1993; Slooter et al., 1998).
Although the influence of APOE genotype has been extensively studied with cognition and brain imaging studies, the risk factor of first-degree family history has not. Results of the REVEAL study (Cupples 2004) highlight the importance of this risk factor, indicating that the risk of developing AD associated with a positive family history was independent of and in addition to the risk attributed to APOE4. In spite of this, little is known about the biologic mechanisms of sporadic AD risk associated with a positive family history, and the relative contributions of family history and APOE genotype on AD risk are unknown. This relationship is important since APOE4 and family history co-occur, and in one study, 45% of adult offspring of AD patients carried the ε4 allele (Sager et al., 2005). An important question is whether the neural effects attributed to APOE are confounded by the influence of other AD risk factors, such as family history.
Functional MRI (fMRI) is useful for detecting hemodynamically coupled neurocognitive brain activity during specific cognitive tasks. Most previous fMRI studies of at-risk subjects have stratified their participants by APOE4 status only. Other studies have examined family history and APOE4 together, but not as separable risk factors, defining the presence of both as high risk and absence of both as low risk (Smith et al., 1999; Fleisher et al., 2005; Smith et al., 2005). No studies to date have evaluated first-degree family history of AD as an explicit risk factor separate from APOE.
In this study, we compared the effects of first-degree family history (defined as having at least one biological parent with AD) and APOE genotype on neurocognitive function using an fMRI encoding paradigm in cognitively normal middle-aged adults. We hypothesized that a family history effect would be observed and that this effect would be independent of APOE genotype.
One hundred, thirty-two subjects underwent fMRI scanning and cognitive testing (Table 1). Sixty-eight participants (mean age 54; SD 6.4) had at least one parent with AD and were recruited from the Wisconsin Registry for Alzheimer’s Prevention (Sager et al., 2005), a longitudinal registry of cognitively normal adults between the ages of 40–65 (at entry) who have at least one parent with sporadic AD (here this group is designated +FH). To verify the diagnosis of AD in the parent, parental medical records were obtained (including autopsy reports when available) and reviewed by a multidisciplinary diagnostic consensus panel. Typically, the clinical work-up and diagnosis in the parent were conducted at the University of Wisconsin Memory Clinics. The onset of AD in the parent was reported to be at age 73 on average. There were no families included with known autosomal dominant mutations. In nine instances the participants were siblings. All subjects in the +FH group underwent baseline neuropsychological evaluations and laboratory tests that included APOE genotyping using PCR and sequencing. 37 (54%) were ε4 positive (ε3/ε4 = 26; ε4/ε4 n = 7, ε2/ε4, n = 4). The remaining 31 were ε4 negative (ε3/ε3 = 24; ε2/ε3 = 6; ε2/ε2 n = 1).
A group of 64 participants (mean age 55; SD 6.5) with no family history of AD (−FH) were recruited from the community and matched to the demographic characteristics of the +FH sample. Absence of first-degree family history of AD was determined through self-report of the participant through phone interview as well as on a detailed medical history questionnaire. To be included in the −FH group, both parents had to survive to at least age 70 and not carry a diagnosis of dementia or exhibit frank symptoms of dementia of any kind. Eleven (17%) of the controls were ε4 positive (ε3/ε4 = 10; ε4/ε4 = 1); 53 were ε4 negative (ε2/ε3 = 6; ε3/ε3 = 47). The demographics of the ε4 positive and negative subgroups are shown in Table 1 along with fMRI task performance, neuropsychological test scores, medication usage, and blood pressure and hemoglobin levels measured the day of the scan. Exclusions for this imaging study included MRI scanner incompatibility, history of major psychiatric disease (e.g., schizophrenia; substance dependence; current or recent major depression) or major medical conditions (e.g., history of neurological disorders including prior head trauma with loss of consciousness, cancer requiring chemotherapy or radiation, insulin dependent diabetes) or abnormal structural MRI or neuropsychological testing as part of study participation. Most psychoactive drugs were excluded, though we did allow low dose SSRIs if stable for more than three months.
The 132 subjects included in the statistical analysis were required to have useable behavioral and imaging data free from artifact or unacceptable motion (movement in the x, y, or z plane > 3 mm). The data from an additional seven subjects were excluded and are not represented in Table 1. These subjects were excluded for neoplasm (a +FH male); cognitive impairment (a +FH male); excessive head motion (a +FH female); scanner error (a +FH female); no APOE result (a −FH male); and questionable or absent task performance data (one +FH male and one −FH female).
The task has been described previously (Johnson et al., 2006) and shown to evoke an MTL response in healthy young and middle aged adults. Further, persons with Mild Cognitive Impairment (MCI) exhibit a significantly reduced response on this task compared to age-matched controls despite comparable task performance. The task involved serial presentation of novel (NV) and previously learned (PL) line drawings obtained from a published normative set (Snodgrass and Vanderwart, 1980). The neurocognitive effect of interest was the cerebral response to NV versus PL items. PL items were acquired in two training sessions 40 minutes and 15 minutes prior to the in-scanner task. The training set consisted of five items with similar frequency and image complexity as the novel items. The training items were presented repeatedly in pseudorandom fashion for 15 exposures in each of the two training trials for a total of 30 exposures to each item. Participants were instructed to view the pictures and try to remember them.
During the fMRI scan, NV or PL items were presented every 3000 ms (2800 ms presentation; 200 ms interstimulus interval) for the duration of the scan, and the task was always to decide if the present item was new or previously learned. There were no periods of rest or cross-hair fixation (see Discussion for rationale), and thus the participants remained in the same cognitive set for the duration of the task. The items were presented as trains of like-events that pseudorandomly varied in number, ranging from single items to five consecutive items of the same type. This variation reduced condition predictability while preserving BOLD efficiency (Liu et al., 2001). Two equivalent forms of the task were sequentially presented (counterbalanced) using the same PL items, but different NV items (45 novel items presented in each form). The task duration for each run was 4 minutes and 42 seconds. Responses were made with a 2-button MR-compatible response device held in the right hand. The software Presentation and a goggle system, set at 800 × 600 from Resonance Technology (Northridge, CA, USA), were used to deliver the stimuli and log participant responses.
The imaging exam was performed with a General Electric 3.0 Tesla scanner (Waukesha, WI). The sequence of scans was as follows: Three-plane localizing scout, higher order shimming (2 iterations), field mapping, fMRI scans, T1-weighted inversion-prepared volume, and fast-recovery fast spin echo T2 weighted axial scans.
A T2* gradient-echo EPI sequence was used. The EPI parameters were as follows: echo time (TE) = 30 ms; repetition time (TR) = 2000 ms; flip angle = 90 degrees; acquisition matrix = 64 × 64 voxels; field of view (FOV) = 240 mm. Thirty sagittal slices of brain were acquired within each TR. Voxel resolution was 3.75 × 3.75 × 5 mm (4mm thick slices with a 1 mm skip). A time course of 141 temporal volume images was collected, of which the initial three volumes were discarded.
To correct for EPI image distortions, 3-D field maps (co-planar with the fMRI slices) were acquired on each subject by measuring the phase of non-EPI gradient echo images at two echo times (7 and 10 ms). Static field inhomogeneity correction was achieved with locally developed software employing a previously published algorithm (Jenkinson, 2003).
The 4-D time-series was motion-corrected (using Statistical Parametric Mapping software; SPM2). The field map from each subject was then applied to the time series. This was followed by spatial normalization into the Montreal standard atlas space using an echoplanar template, re-sampling to 2 mm isotropic voxels and spatial smoothing to 8 mm. To derive single subject activations, the conditions were convolved with a canonical hemodynamic response function. High frequency signal filtering (cutoff = 1/128 Hz) was used. Non-sphericity due to temporal autocorrelation was estimated with an autoregressive (AR1) model. The contrast NV > PL was computed for each subject and subsequently entered into a second-level random-effects 2 × 2 factorial ANOVA to examine the group interactions and main effects of family history and APOE4 status. The omnibus F statistic was computed and subsequent analyses were restricted only to regions where this was significant pFWE<.05, corresponding to an F statistic of 8.88. This procedure reduces the risk of false positive errors from the thousands of voxel-wise tests that are performed by reducing the search region to those voxels where at least one of the four groups was significantly different from zero. We then tested the group t contrasts of −FH > +FH and its opposite (both comprising the main effect of FH factor); −ε4 < +ε4 and −ε4 > +ε4 (comprising the main effect of ε4 factor); and the interaction between factors. We also tested the effect of FH on only ε4 positive carriers. All of these contrasts were corrected for multiple comparisons using the False Discovery Rate (FDR) method (Genovese et al., 2002) pFDR < .005 and a cluster threshold of at least 100 contiguous 2 × 2 × 2 mm voxels (greater than 0.8 cubic cm). Similar ANCOVA models were subsequently tested for reaction time and gender to determine their influence on the findings.
Axial T1 and T2 weighted images were acquired after the functional runs. A 3D IR-prepped fast gradient echo pulse sequence provided high-resolution T1-weighted structural images with the following parameters: inversion time = 600 ms, fast gradient echo read-out with TR/TE/flip = 9 ms/1.8 ms/20°; acquisition matrix = 256 × 192 × 124 (interpolated to 256 × 256 × 124); field of view = 240 mm; slice thickness = 1.2 mm (124 slices); ± 16 kHz receiver bandwidth.
A Fast Recovery Fast Spin Echo 2D T2-weighted axial sequence was also acquired with the same start and stop locations as the T1 weighted images. The parameters were: field of view = 240 mm, matrix 256 × 256, TR = 9000 ms, TE = 93 ms, flip angle = 90. Seventy slices were acquired; slice thickness = 1.7 mm with 0.3 mm skip. An experienced neuroradiologist examined all images prior to the analysis for clinical evidence of any neurovascular disease or structural abnormality that would exclude the subjects from the analysis.
The T1 volume was subsequently used for voxel-based morphometry (VBM) to determine whether there were volumetric differences between groups that might account for any observed fMRI differences. The VBM procedure utilized a standard approach (Good et al., 2001) that included optimized normalization to standard atlas space (and resampled at 2 mm isotropic voxels), modulation of the normalized image by the Jacobian determinants in order to preserve volume information, followed by spatial smoothing to 12 mm. The Gaussian smoothing function differed from that of the functional images because the intrinsic smoothness of the high resolution structural scans was less than that of the functional scans. A comparable 2 × 2 statistical design was used to determine whether there were gray matter differences at a threshold of pFDR < .05 corrected for multiple comparisons.
Table 1 includes demographic, neuropsychological, behavioral and laboratory data. The groups did not differ on age, education, cognitive test performance or fMRI task accuracy. Reaction time to PL items was equivalent between groups; however, reaction time to NV items was slightly slower (~100 ms on average) in the +FH subjects; this main effect reached statistical significance and NV reaction time was treated as a covariate in a follow up ANCOVA. Chi square analyses of medication usage indicated that the groups were equivalent on the proportion taking non-steroidal anti-inflammatory and cholesterol-lowering drugs. However, use of hormone replacement therapy (HT) in women, antihypertensives, and antidepressants were higher in one or both of the +FH groups. We therefore computed follow-up analyses to determine whether medications might have affected the overall findings. Gender proportion was different between subgroups (ChiSquare=8.37, p=.03) and was treated as a covariate in a follow up ANCOVA analysis.
Although not part of the planned comparisons, the results from the −FH and +FH groups are presented separately in the first two rows of Figure 1 and Table 2 to show qualitatively where each group activated. These maps were masked with the omnibus F as described above and further constrained to pFWE< .05 (t=4.11; 100 voxel extent threshold). The reverse contrasts are also reported in Figure 1. In the −FH group there was a single region in the precuneus more active during PL>NV (x,y,z 10, 68, 40; t=5.30, pFWE =.001; 253 voxels) . In the +FH group there were two clusters involving the right lateral parietal lobe (44,−54,32; t=6.61; pFWE<.001; 1137 voxels) and right precuneus (6,−74,34; t=5.73; pFWE<.001; 589 voxels) where the PL>NV contrast was significant.
The main effect of FH is shown in the third row of Figure 1, and in Table 1. There was a robust main effect of first-degree FH where −FH activated more than +FH in the ventral and mesial temporal lobe including the hippocampus bilaterally (right more than left). The contrast +FH > −FH revealed no significant voxels.
There was no main effect of APOE4 status, meaning that no voxels survived the analysis (in either direction). An F-test for the interaction of family history and ApoE was not significant at the prespecified threshold.
To further describe the response in our hypothesized areas of interest, we selected the locations of maximal activity in the right and left hippocampi at 34, −22, −18 (right) and −26, −20, −16 (left) and plotted the signal values. The result of the plot for the right hippocampus location is shown in Figure 2. The −FH, +ε4 group exhibited the greatest signal change followed by the −FH, −ε4 group. The +FH, +ε4 group activated the least. Although this interaction was not significant at the prespecified FDR corrected threshold and voxel extent threshold, it is significant at an uncorrected threshold for the right hippocampus (F [3,128] = 12.6, p = .001) as well as the left (F [3,128] = 4.2, p = .043) and is presented here descriptively.
To address our hypothesis that the ε4 effect would be influenced by family history, we examined the contrast of −FH, +ε4 vs +FH, +ε4. This contrast resulted in significantly greater signal change in the −FH, ε4+ group in the right ventral temporal lobe and hippocampus as well as the left hippocampus (see Figure 3 and Table 3). There were no significant findings in the reverse contrast.
The four cells of the 2×2 design differed in gender composition (ranging from 45% to 77% women; see Table 1); therefore the same 2×2 model was set up as an ANCOVA with gender as the covariate to determine its influence on the results. Using a contrast that weighted the covariate column with 1, there were no regions where gender significantly influenced the result even at a liberal uncorrected threshold of p<.001. Further, the main effect of FH, and the APOE4 by FH effect in the presence of the covariate were each highly similar in magnitude and spatial pattern to the original model, leading us to conclude that gender composition did not have a significant effect on these results.
The −FH subjects had slightly faster reaction times to NV items. To determine whether this was influencing the result a separate ANCOVA was conducted with NV reaction time as a covariate. As with gender, the pattern of findings remained highly significant (pFDR<.01) in the presence of the reaction time covariate, leading us to conclude that reaction time was not accounting for our results.
Common medications used by participants are described in Table 1. In particular the +FH, +ε4 group were taking more antidepressants (10 subjects or 27% of the group) and the +FH, −ε4 group were taking more antihypertensives (8 subjects; 26%). However, scores on a depression scale and mean systolic blood pressure were similar across all groups, suggesting these factors were likely not affecting our results. Nevertheless, we conducted two additional analyses to test these possible effects. In the first analysis, we randomly excluded 8 of 10 subjects in the +FH, +ε4 group who were taking antidepressants (reducing the proportion to 7%) in order to equate the proportion taking this class of drug across all four groups. The analysis revealed the same general pattern: there was a substantial main effect of family history in the mesial and ventral temporal lobes (FDR p < .05), and there continued to be no effect of APOE. The second analysis randomly excluded 6 of 8 subjects in the +FH,−ε4 group who were taking blood pressure lowering medications (reducing the proportion on blood pressure lowering medication in that group from 26% to 8%). The general pattern of findings was again highly similar: the family history main effect was quite robust (surpassing pFDR < .05 threshold) while the ε4 main effect across groups remained non significant.
To ensure the fMRI effect was not due to regional differences in brain atrophy, a voxel-based analysis of gray matter volume was performed. Using the same statistical design and threshold of pFDR < .05, there were no significant differences in gray matter volume anywhere in the brain. For comparison with the fMRI results, the VBM statistics are shown in Table 2 and Table 3 at each of the fMRI maxima.
In this study, we found that family history of AD was an important predictor of MTL activation independent of APOE genotype. The hippocampus exhibited significantly less activation in ε4 carriers with a family history which is similar to findings in other studies in persons with MCI (Small et al., 1999; Machulda et al., 2003; Johnson et al., 2004; Johnson et al., 2006). This finding that asymptomatic ε4 carriers with a parental history of AD are exhibiting reduced signal change in regions of the brain known to be sites of early AD pathology suggests that this is a particularly high risk group for developing AD. This bilateral main effect occurred in the absence of meaningful differences in neuropsychological test results, task accuracy, and regional differences in brain volume, and the results were not attributable to group differences in gender composition, task reaction time or medication usage. The interaction between family history and APOE was not significant at a pre-specified stringent threshold. However, more specific to our hypothesis, a direct contrast of ε4 carriers with and without family history of AD indicated that −FH ε4 carriers exhibited significantly more signal change in the mesial and ventral temporal lobe. Prior fMRI studies do not provide a consensus on whether AD-vulnerable regions, such as the MTL, show a diminished response associated with risk factors (Small et al., 1999; Smith et al., 1999; Johnson et al., 2004; Smith et al., 2005; Trivedi et al., 2006) or a paradoxical increased response (Bookheimer et al., 2000; Bondi et al., 2005; Dickerson et al., 2005; Fleisher et al., 2005). It is intriguing that the −FH ε4 carriers in this study exhibited the greatest signal change in the MTL. This aspect of our results replicates prior studies that have reported an increased response (Bookheimer et al., 2000; Bondi et al., 2005), and suggests that modeling FH may help explain discrepancies in the literature on whether APOE confers an increased or decreased fMRI response in people at risk for AD.
The ventral temporal lobe (fusiform gyrus) also exhibited a significant main effect of family history. The region is involved in object identification and has been shown in several studies to be more responsive to novel items in nonhuman primates (e.g. Ringo, 1996) and in humans (Detre et al., 1998; Rand-Giovannetti et al., 2006). The region has been found to be one of several regions that are atrophic in MCI subjects who later convert to AD (Chetelat et al., 2005). In AD patients undergoing a face encoding task, the fusiform was less active than controls, but improved with ten weeks of cholinesterase-inhibitor treatment (Kircher et al., 2005).
The current findings in persons with risk of AD are consistent with PET studies supporting the concept that reduced glucose metabolism may lead to reduced neuronal activation in the MTL during fMRI tasks. CMRgl is coupled to regional cerebral blood flow (rCBF) (Sokoloff, 1984, 1984), and both CMRgl (Mosconi et al., 2004) and rCBF (Heeger and Ress, 2002; Logothetis and Wandell, 2004) are coupled to the neural response. More research is needed to determine the relationship between CMRgl in patient groups and task-related changes in the fMRI BOLD signal. In the present analyses the canonical hemodynamic response function (HRF) was used to estimate the neural response. Others have shown that the HRF is reduced in older adults (D'Esposito et al., 1999) and delayed in persons with dementia (Rombouts et al., 2005). Although the middle-aged subjects in this report were cognitively normal, differences in the shape of the HRF as a function of group may have contributed to the effects observed here. The structural integrity of the MTL may also affect fMRI signal. MTL volumetric APOE4 effects have been observed in older adults (Moffat et al., 2000; Lemaitre et al., 2005), and trends have been reported in middle-aged adults at risk for AD (Reiman et al., 1998). Our VBM result in the MTL also suggests a non-significant trend. Combined ROI analyses of MTL structure and function (Dickerson et al., 2004; Dickerson et al., 2005) may help further delineate the effect of atrophy on MTL fMRI signal change.
The findings of the present study suggest that an unknown factor or factors embodied in a family history of AD may be influencing the expression of ε4 on brain function. This possibility should not be surprising, given the accumulating evidence that the risk of developing AD can be influenced by many factors (Cupples et al., 2004) including newly discovered gene-gene interactions (Borroni et al., 2004; Bernardini et al., 2005; Dunckley et al., 2005; Lambert et al., 2005; Papassotiropoulos et al., 2005), gene-race interactions (Green et al., 2002; Gureje et al., 2006), as well as interactions with other medical (Evans et al., 2004), and lifestyle/environmental factors (Mutter et al., 2004; Bird, 2005; Salerno-Kennedy and Cashman, 2005), all of which contribute to disease pathogenesis in poorly understood ways. More work is needed to understand the factors in family history that confer AD risk and their influence on brain structure and function.
The present study examined the relative cerebral response to novel items versus previously learned items using an easily performed task based on a well-known paradigm (Stern et al., 1996; Tulving et al., 1996; Detre et al., 1998; Saykin et al., 1999; Johnson et al., 2006). This approach assumes the MTL, a region vulnerable to AD pathology, is involved in encoding novel episodic information (Tulving et al., 1994). The fMRI paradigm for this study consisted of variable length epochs and changes in condition were not announced to the participant. The participants thus remained in a discriminatory (new vs old) cognitive set for the duration of the task. The variable length design had many of the advantages of boxcar-style designs where a steady cognitive state is assumed, including greater efficiency for detecting the BOLD signal (Liu et al., 2001), and greater statistical power (Henson, 2003). The variability in epoch length reduced cognitive predictability. We did not use a low-level baseline control task such as cross-hair fixation. Rather, we used previously encoded items as the baseline condition. Stark and Squire (2001) found that MTL signal change was significantly greater when an active task was used as a baseline rather than rest. Furthermore, the posterior cingulate and hippocampus exhibit correlated activity during passive fMRI tasks such as cross-hair fixation relative to more goal directed tasks (Greicius and Menon, 2004; Greicius et al., 2004). Thus, the assumption that rest or other low level baseline is psychologically and physiologically neutral is likely not true for regions such as the hippocampus, and for this reason, we avoided a resting baseline. Figure 1 indicates that the medial parietal lobe was more active to the PL baseline than NV items, consistent with prior research on recognition (Buckner et al., 2005; Wagner et al., 2005). However, the groups did not differ from each other in this region. We have not attempted to characterize this region in great detail because the PL condition is limited to only five items that repeat, making it less than ideal for studying recognition effects. Other studies in our laboratory are ongoing to investigate this region in greater detail.
There are limitations to this study. First, we relied on a consensus diagnostic conference to review parental medical records. Although consensus conference diagnoses were based on established clinical diagnostic criteria for AD, only 10% of the family history cohort had parents with autopsy-proven AD. As a result, our family history cohort may have been contaminated with subjects whose parents did not have AD. Second, the control group may have contained persons whose parents may eventually develop AD. We tried to minimize this by requiring parents to have survived to at least the age of 70 without evidence of a dementing illness. Nevertheless, we cannot exclude the possibility that the control cohort may have contained persons whose parents may eventually develop AD. Both of these possibilities would have had the effect of diminishing the robustness of our findings and illustrate the need for diagnostic biomarkers in AD research. Another potential limitation is the unequal sample sizes in the ANOVA. The cell sizes were proportionate to the populations from which they were drawn, however the −FH +ε4 group only contained 11 subjects rendering some of our analyses vulnerable to both low statistical power to detect effects (although effects were indeed observed), and undue influence from outliers. We have shown the range of the data in Figure 2 and these box plots indicate that the within-group range of signal change was equivalent across groups and our inspection of the data revealed no outliers in the group of 11. This group was small in size because only approximately 15% of the normal population are ε4 positive, while approximately half of people with a first degree relative with AD are ε4 positive (Sager et al., 2005).
We found that family history of AD was an important predictor of MTL activation independent of APOE genotype in cognitively normal people at risk for AD. This finding may help reconcile the conflicting literature on the fMRI response in preclinical AD. Our findings underscore the importance of evaluating other factors, like family history, that may interact with APOE genotype in conferring risk of AD. To date, the factor or factors that impart AD risk to family history are poorly understood (Bertram et al., 2005; Tanzi and Bertram, 2005). A recent study found that monozygotic twins have a concordance rate for AD of 59%, highlighting the fact that genetics alone do not completely account for sporadic AD and that other variables are involved in AD pathogenesis (Gatz et al., 2005).
This study has implications for future research. First, our findings need to be replicated and expanded upon. Imaging studies that measure neurofibrillary tangles or amyloid deposition (Klunk et al., 2004) glucose uptake (Reiman et al., 2004) or resting perfusion (Johnson et al., 2005) are required to better delineate the neurobiology of AD risk conferred by a family history of AD. Second, although persons with risk factors for AD have always been important volunteers for AD research, a systematic study of the biological and neurocognitive changes occurring in middle-aged first-degree relatives is needed to characterize the pre-clinical course of disease. Finally, longitudinal cohort studies are needed to determine the genetic, environmental and lifestyle factors that contribute to the risk of AD associated with family history, and to determine whether the findings we report are predictive of subsequent cognitive decline and eventual development of MCI or AD.
R01 AG21155 (SCJ) supported this study. We thank Tim Hess, MS, Andrew Alexander, Ph.D., Howard Rowley, M.D. for their assistance in accomplishing this work.