Abstract

Advanced age and vascular risk negatively affect episodic memory. The hippocampus (HC) is a complex structure, and little is known about the roles of different HC regions in age-related memory declines. Using data from an ongoing longitudinal study, we investigated whether memory functions are related to volumes of specific HC subregions (CA1-2, CA3-4/dentate gyrus, and subiculum). Furthermore, we inquired if arterial hypertension, a common age-related vascular risk factor, modifies age-related differences in HC regional volumes, concurrent memory performance, and improvement in memory over multiple administrations. Healthy adults (n = 49, 52–82 years old) completed associative recognition and free recall tasks. In grouped path models, covariance structures differed between hypertensive and normotensive participants. Whereas larger CA3-4/dentate gyrus volumes predicted greater improvement in associative memory over repeated tests regardless of vascular risk, CA1-2 volumes were associated with improvement in noun recall only in hypertensive participants. Only among hypertensive participants, CA1-2 volumes negatively related to age and CA3-4/dentate gyrus and CA1-2 volumes were associated with performance at the last measurement occasion. These findings suggest that relatively small regions of the HC may play a role in age-related memory declines and that vascular risk factors associated with advanced age may modify that relationship.

INTRODUCTION

Advanced age is accompanied by difficulties in recall of relational and contextual information (Old & Naveh-Benjamin, 2008; Spencer & Raz, 1995), and brain substrates of episodic memory (e.g., the hippocampal formation, Squire & Zola, 1996) deteriorate with age as well (see Raz & Rodrigue, 2006, for a review). Thus, it is plausible that age differences in memory would be related to differences in structural integrity of the hippocampus. Yet, the findings from correlational studies of hippocampal volume and memory in healthy adults are inconsistent and contradictory (see Van Petten, 2004, for a review). Although the reasons for the discrepant reports from MRI studies on the hippocampus and memory may be manifold, several of them stand out, including relative coarseness of anatomical and cognitive measurements, differences in sample composition, and failure to account for the influence of common vascular risk factors.

To date, most in vivo neuroanatomical studies of hippocampal involvement in human memory have been restricted to the assessment of the total structure. However, the hippocampus is a complex and heterogeneous entity with significant cytoarchitectonic variations across multiple regions (Witter & Amaral, 1995). Thus, aggregating discrepant anatomical substrates may obscure the true associations between structure and function.

The hippocampus is sensitive to elevated vascular risk and shows accelerated age-related shrinkage in persons with history of hypertension (Raz et al., 2005). Thus, vascular risk factors may contribute to age differences in hippocampal volume and memory as well as to the association between them. Moreover, some of the hippocampal subregions demonstrate differential vulnerability to various vascular insults and risk factors. For instance, sector 1 of the cornu ammonis (CA1) is particularly sensitive to hypoxia, ischemia, and hypertension (Sugawara, Lewen, Noshita, Gasche, & Chan, 2002; Sabbatini, Strocchi, Vitaioli, & Amenta, 2000; Nitatori et al., 1995; Suyama, 1992; Kirino & Sirno, 1984; see Schmidt-Kastner & Freund, 1991, for a review), and its role in memory may be differentially influenced by vascular risk and vascular disease (Yonelinas et al., 2002).

Vascular risk may play an important role in determining the relations between regional hippocampal volumes and memory performance. Declines in brain structure and selected cognitive abilities are exacerbated by age-associated vascular risk factors such as essential hypertension (Brady, Spiro, & Gaziano, 2005; Singhmanoux & Marmot, 2005; Waldstein, Brown, Maier, & Katzel, 2005; Waldstein & Katzel, 2001; Harrington, Saxby, McKeith, Wesnes, & Ford, 2000), and the influence of vascular risk on brain and cognition varies across brain regions and cognitive domains (see Raz & Kennedy, 2009; Birns & Kaira, 2008, for reviews). Elevated arterial blood pressure has been linked to impaired psychomotor speed as well as poor fluid reasoning, working memory, and executive functions (Semplicini et al., 2010; Waldstein et al., 2005; Kuo et al., 2004; Raz, Rodrigue, & Acker, 2003; Saxby, Harrington, McKeith, Wesnes, & Ford, 2003; Knopman et al., 2001; Harrington et al., 2000; Elias, Robbins, Elias, & Streeten, 1998). Hypertension is also associated with reduced performance on various measures of episodic memory, including immediate and delayed recognition and free recall (Hannesdottir et al., 2009; Kuo et al., 2004; Saxby et al., 2003; Harrington et al., 2000). However, the neuroanatomical correlates of such associations remain unclear.

Recent refinement of in vivo MRI methods using high in-plane resolution enabled reliable demarcation of hippocampal subregions in humans. Studies employing such methods revealed that older adults have smaller CA1-2 volumes than their younger counterparts, with smaller or negligible differences in other hippocampal regions (Mueller, Chao, Berman, & Weiner, 2011; Mueller & Weiner, 2009; Mueller et al., 2007). Because even ostensibly healthy aging is accompanied by elevation of blood pressure (Wills et al., 2011), it is possible that the observed age differences reflect an admixture of participants with hypertension whose CA1-2 volumes are smaller than those of their normotensive counterparts (Shing et al., 2011).

In addition to the refinement of neuroanatomical measures, investigations of structural substrates of age differences in memory may benefit from more fine-grained measures for memory assessment. Memory is not a uniform construct (Moscovitch, Nadel, Winocur, Gilboa, & Rosenbaum, 2006), and recent studies that used high-resolution measures of hippocampal anatomy indicate that regional volumetric differences may be associated with different types of memory. For example, in older adults, larger volume of the CA3-4 sector and the dentate gyrus (DG) was linked to better verbal recall and associative recognition after a short delay (Mueller et al., 2011; Shing et al., 2011), whereas greater CA1-2 volume has been associated with improved consolidation and retrieval following a relatively long delay (Mueller et al., 2011).

In this study, our goal was to address the outlined limitations of the extant literature. In a sample of healthy middle-aged and older adults, we measured the volumes of three hippocampal regions, assessed two types of memory (association and free recall), and took into account the contribution of a common age-related vascular risk factor, essential hypertension. Moreover, we took advantage of the availability of repeated assessments of memory separated by about 2 years. Thus, in addition to baseline performance, we were able to gauge the gains in memory scores after repeated exposure to the same material. Because these benefits are independent of age (Salthouse, 2013), we sought to investigate the hippocampal subregions as possible neuroanatomical correlates of longitudinal effects of repeated testing. We used a structural equations approach (path analysis) to examine the relationships among age, hippocampal subfield volumes, and longitudinal change in two types of memory—cued associative recognition and verbal free recall—while taking into account the moderating effects of vascular risk. We hypothesized that, in a sample of healthy middle-aged and older adults, history of hypertension modifies the relationships among age, memory, and subregional hippocampal volumes, as evidenced by different underlying variance–covariance structures. We hypothesized that, in accord with Mueller et al. (2011) and Shing et al. (2011), larger volume of the CA3-DG region would predict better association memory and that advanced age would be associated with poorer memory and with smaller volumes of the CA1-2 subfields. Moreover, we expected that, because of differential sensitivity of the CA1-2 subfield, age-related differences in volume of that region would be associated with poorer memory among hypertensive participants.

METHODS

Participants

The sample was drawn from an ongoing longitudinal study of brain and cognitive aging in the Detroit Metropolitan area. Participants completed a 66-item health questionnaire that included questions regarding current or prior diagnosis and treatment of essential hypertension and the names of medications prescribed for the treatment of hypertension. The questionnaire also screened for neurological, psychiatric, or cardiovascular disease; cancer except nonmalignant skin cancer; head injury accompanied by loss of consciousness for over 5 min; diagnosed or treated thyroid disorder; diabetes; treatment for drug and alcohol abuse; or taking more than three alcoholic drinks per day. Persons who reported use of anticonvulsive, anxiolytic, antipsychotic, and antidepressant medications were not enrolled in the study. All participants were native English speakers with at least a high school diploma or equivalency (mean education = 15.41, SD = 2.25 years), and all were right-handed (Oldfield, 1971).

The participants completed a questionnaire to rule out current symptoms of depression (CES-D; Radloff, 1977; cutoff = 15), and an experimenter administered the Mini Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975; cutoff = 26) to screen for cognitive impairment. The experimenters screened all participants for near, far, and color vision problems (Optec 2000 Vision Tester, Stereo Optical Co., Inc., Chicago, IL) and speech-range hearing deficits (MA27 Screening Audiometer, Maico Diagnostics, Eden Prairie, MN). Trained laboratory staff measured blood pressure using an auscultatory method using diastole phase V for the identification of diastolic pressure (Pickering et al., 2005). Blood pressure was measured with the participant comfortably seated in a quiet room. The measurements were taken on 3 separate days, normally 1–2 weeks apart, and values were averaged across the three occasions.

All participants provided informed consent in accord with the University Human Investigation Committee guidelines. The sample consisted of 49 healthy adults, (13 men, 36 women) ranging in age from 52 to 82 years (mean age = 60.84 years, SD = 9.59 years). Men and women did not differ in mean age, years of education, MMSE scores, or systolic blood pressure (t < 1, ns for all), but men had significantly higher diastolic blood pressure (M = 79.58 mmHg, SD = 6.52 mmHg) than women did (M = 75.23 mmHg, SD = 6.54 mmHg; t(47) = 2.057, p < .05).

The sample included 28 normotensive adults and 21 who reported medically treated, essential hypertension. The proportion of men and women did not differ by hypertension status (χ2 = 0.873, p > .1). Normotensive and hypertensive groups did not differ in mean age, education, MMSE score, or systolic blood pressure (Table 1). Although mean diastolic blood pressure was higher in the hypertensive group than in the normotensive group, it was well under the diagnostic cutoff, indicating sufficient control (Chobanian et al., 2003). The sample originally included three additional participants, but because of motion artifacts on their MRI images, these data were not included in the analyses. All three were normotensive and did not differ from the included sample in mean age (t(2.3) = 0.909, p > .1), years of education (t(2.1) = 0.627, p > .1), MMSE scores (t(2.1) = 1.840, p > .1), or systolic and diastolic blood pressures (for both ts < 1, ps > .1). The sample included 21 participants who were 52–60 years old (15 women), 12 who were 61–70 years old (10 women), 12 who were 71–80 years old (seven women), and four women who were 81–82 years old.

Table 1. 

Sample Demographic Characteristics

Variable
Normotensive, Mean ± SD
Hypertensive, Mean ± SD
t
p
Age (years) 65.04 ± 9.27 66.05 ± 10.13 −0.364 .718 
Years of education 15.89 ± 2.18 14.76 ± 2.23 1.777 .082 
MMSE score 29.14 ± 1.11 28.76 ± 0.94 1.264 .213 
Systolic BP (mmHg) 124.21 ± 12.38 127.53 ± 10.51 −0.991 .327 
Diastolic BP (mmHg) 74.11 ± 6.02 79.42 ± 6.60 −2.931 .005* 
Duration of BP treatment (years)  7.56 ± 1.60   
Memory for Names: immediate 41.25 ± 11.56 38.14 ± 10.32 0.974 .335 
Memory for Names: delayed 24.68 ± 10.36 22.29 ± 9.72 0.821 .416 
Noun list recall: immediate 6.07 ± 2.71 6.00 ± 3.07 0.086 .932 
Noun list recall: delayed 4.25 ± 2.72 4.35 ± 2.56 −0.129 .898 
EC volume (mm3529.74 ± 65.98 516.21 ± 97.81 0.579 .566 
SUB volume (mm3193.99 ± 25.19 186.76 ± 31.60 0.891 .378 
CA1-2 volume (mm3336.28 ± 41.81 316.45 ± 35.82 1.745 .088 
CA3-4/DG volume (mm3213.74 ± 33.16 212.92 ± 24.89 0.095 .925 
Variable
Normotensive, Mean ± SD
Hypertensive, Mean ± SD
t
p
Age (years) 65.04 ± 9.27 66.05 ± 10.13 −0.364 .718 
Years of education 15.89 ± 2.18 14.76 ± 2.23 1.777 .082 
MMSE score 29.14 ± 1.11 28.76 ± 0.94 1.264 .213 
Systolic BP (mmHg) 124.21 ± 12.38 127.53 ± 10.51 −0.991 .327 
Diastolic BP (mmHg) 74.11 ± 6.02 79.42 ± 6.60 −2.931 .005* 
Duration of BP treatment (years)  7.56 ± 1.60   
Memory for Names: immediate 41.25 ± 11.56 38.14 ± 10.32 0.974 .335 
Memory for Names: delayed 24.68 ± 10.36 22.29 ± 9.72 0.821 .416 
Noun list recall: immediate 6.07 ± 2.71 6.00 ± 3.07 0.086 .932 
Noun list recall: delayed 4.25 ± 2.72 4.35 ± 2.56 −0.129 .898 
EC volume (mm3529.74 ± 65.98 516.21 ± 97.81 0.579 .566 
SUB volume (mm3193.99 ± 25.19 186.76 ± 31.60 0.891 .378 
CA1-2 volume (mm3336.28 ± 41.81 316.45 ± 35.82 1.745 .088 
CA3-4/DG volume (mm3213.74 ± 33.16 212.92 ± 24.89 0.095 .925 

p < .05.

Although 197 participants aged 47 years and older completed the baseline memory assessments, due to MRI hardware upgrades only 49 members of the current sample underwent high-resolution HC imaging on the 3-T magnet in addition to completing three waves of memory assessment. In comparison with the 148 who did not complete longitudinal testing and high-resolution HC MRI at Time 3, the present sample did not differ with age (t = 0.407, p = .684), sex composition (χ2 = 0.89, p = .345), years of education (t = 0.529, p = .597), proportion of hypertensive participants (χ2 = 0.351, p = .554), or mean systolic and diastolic blood pressures (for both: ts < 1.0, ps > .5). MMSE scores were marginally but significantly higher than in the total baseline sample (t = 2.129, p < .05).

MRI Acquisition and Processing

MRI Acquisition

The following sequences relevant to this study were acquired as part of a 1-hr protocol on a 3-T Siemens Verio (Siemens Medical AG, Erlangen, Germany) full-body magnet with a 12-channel head coil. To measure the intracranial volume (ICV), we acquired a high-resolution T1-weighted magnetization prepared rapid gradient-echo sequence with the following parameters: repetition time = 1680 msec, echo time = 3.51 msec, inversion time = 900 msec, flip angle = 9.0°, pixel bandwidth = 180 Hz/pixel, GRAPPA acceleration factor PE = 2, and voxel size = 0.67 mm × 0.67 mm × 1.34 mm. For regional hippocampal measures, we acquired a high-resolution PD-weighted turbo spin echo sequence in the coronal plane, oblique to the long axis of the hippocampus, with the following parameters: echo time = 17 msec, repetition time = 7150 msec, flip angle = 120°, pixel bandwidth = 96 Hz/pixel, turbo factor 11, voxel size = 0.4 mm × 0.4 mm × 2.0 mm, field of view = 280 × 512 mm, 30 slices; we limited the field of view to a smaller 2-D cross-section to allow for faster acquisition.

Postprocessing and Manual Demarcation of Anatomical Regions

Rules for tracing the hippocampal subfields and entorhinal cortex (EC) were adapted from Shing et al. (2011) as modified from Mueller & Weiner (2009) and Mueller et al. (2007), with one difference: The border between subiculum (SUB) and CA1-2 followed the tissue contrast gradient rather than a vertical line (see Figure 1 for an example of the ROI tracings). Using Analyze 10.0 software (Mayo Clinic, Rochester, MN), two independent raters (A.R.B. and A.M.D.) manually demarcated regional boundaries with a stylus on a 21-in. digitizing tablet (Wacom Cintiq, Vancouver, WA). Interrater reliability was confirmed by an intraclass correlation coefficient for independent raters [ICC(2); Shrout & Fleiss, 1979] of at least 0.90 for a bilateral total volume for each region. Regions included EC, SUB, CA1-2, CA3-4, and DG.

Figure 1. 

Regional measurements of the SUB (in white), CA1-2 (CA1 in purple), CA3-4, and DG (in yellow), and EC (in red) are shown. Image is in radiological orientation.

Figure 1. 

Regional measurements of the SUB (in white), CA1-2 (CA1 in purple), CA3-4, and DG (in yellow), and EC (in red) are shown. Image is in radiological orientation.

To improve visualization, the image intensities were inverted to mimic an inversion-recovery T1-weighted image. However, given the limited visualization of anatomical boundaries between subfields, CA3 and DG were collapsed into a single region, as were CA1 and 2 (see Figure 1). All regions were traced in both hemispheres. Ranges were allowed to differ by starting the slice based on anatomical hemispheric differences. To ensure separation between HC and amygdala, the HC subfield range began with the slice on which the head of the HC was no longer visible.

Entorhinal cortex

The EC was traced on six contiguous slices, beginning with five slices anterior to the starting slice of the hippocampal subfield range. The end of the SUB defined the superior medial boundary, and the opening of the collateral sulcus was the inferior lateral boundary. Reliability for this region was ICC(2) = 0.99.

SUB

The SUB was traced on three slices beginning on the starting slice. The region was traced from the end of the CA1 to the dorsal EC boundary, which was determined by extending a diagonal line from the medial border of the DG. Reliability for this region was ICC(2) = 0.93.

CA1-2

The CA1-2 (CA1) region was traced on three slices beginning on the starting slice. The boundary circled the CA3 and DG, extending medially from the dorsal border to the SUB. Reliability for this region was ICC(2) = 0.91.

CA3/4 and DG

This region (DG) was traced on the three contiguous slices with the other subfields. The ovoid region was traced within the hyperintense border between CA1-2. Reliability for this regions was ICC(2) = 0.93.

ICV

The ICV was calculated using the brain extraction tool (BET; Smith, 2002) in FSL 4.1 using the T1-weighted images. To minimize the inclusion of nonbrain tissue such as eyes, periorbital fat, or neck muscle and fat, we first applied standard-space masking (starndard_space_roi; Keihaninejad et al., 2010). Cranium was demarcated in BET by applying a fractional intensity threshold of 0.2, without gradient; the method used the “−A” flag for the estimation of the skull from the masked image with the betsurf option (Jenkinson, Pechaud, & Smith, 2005). ICV values were sampled from the outer skull mask output by betsurf, and an experienced operator visually inspected the results. These settings were determined following the comparison of hand-tuned BET output on a subsample.

Regional volumes were corrected for ICV via a linear equation: volumeadj = volumerawib(ICVi − mean ICV), where volumeadj is the adjusted regional volume, volumerawi is the original volume for an individual, b is the slope of the ROI volume regressed on ICV, and mean ICV is the sample mean of ICV. In this correction, we multiplied the ICV values by 1000 to equalize the scales of HC and ICV.

Tests of Episodic Memory

Free Recall: Word List

Stimuli

Test stimuli, combined in lists of 16 nouns, originated from the Toronto Word Pool (Friendly, Franklin, Hoffman, & Rubin, 1982). Stimuli lists for each participant were randomly drawn from a pool of six lists. The same word list was presented at all three longitudinal occasions to test short-delay free recall (SDFR) with repeat exposure. At each longitudinal administration, additional (nonrepeated) noun lists were also presented for testing both SDFR and long-delay free recall (LDFR).

Procedure

A computer screen displayed the list of 16 nouns, serially presented for 3 sec each, with a 200-msec ISI; the total study time for each word list was 51 sec. The experimenter instructed participants to study the words and try to remember them. After studying, a 1-min distractor task required participants to audibly count backward by threes from a random 900 number to minimize rehearsal. After a 1-min delay, the experimenter asked the participant to recall and articulate as many words as possible from the study, regardless of order. The experimenter recorded the SDFR responses. For each participant, the same SDFR list used at baseline was repeatedly administered at every longitudinal measurement occasion. In addition to the repeated SDFR list, longitudinal administrations also presented a new list of 16 nouns that had not been previously administered. For the second nonrepeated list, participants were asked to freely recall as many words as possible following both a short delay of 1 min and a longer delay of 20 min. Participants were not informed of the recall delays. The total number of words correctly recalled served as the index of performance for both SDFR and LDFR tests.

Recognition: Picture–Name Associations

Participants completed the Memory for Names subtest of the Woodcock–Johnson Psychoeducational Battery-Revised (Woodcock & Johnson, 1989). The experimenter serially presented novel visual stimuli, pictures of “space creatures,” and stated the creature's name consisting of one- and two-syllable nonsense stimuli. After the presentation of each new picture, the experimenter presented a page with multiple space creatures and asked the participant to point to each previously studied item after the experimenter stated its name; the experimenter provided the correct answer following incorrect responses. The maximum score possible was 72 correct responses. After a 20-min delay, the experimenter serially presented 12 pages, each containing 12 space creatures, and instructed the participant to point to a specific creature using its previously learned name. At delayed testing, one trial per page was presented for 36 trials, with no feedback following the responses. The two performance indices were the total number of correct responses for immediate (NamesImmediate) and delayed cued (NamesLDCR) associative memory tests. Both immediate and delayed tests have estimated reliabilities of .91 (Woodcock & Johnson, 1989).

Participants returned for a second testing occasion (T2), a mean of 24.76 months following initial testing (T1). However, hypertensive participants (mean = 25.26, SD = 1.32 months) had a significantly longer mean delay between T1 and T2 than normotensive participants (mean = 24.41, SD = 0.94 months; t(45) = 2.60, p < .05). The mean delay between T2 and the third testing occasion (T3) was 31.31 months and did not differ between the groups (normotensive: mean = 32.21, SD = 5.31 months; hypertensive: mean = 29.97, SD = 4.61 months; t(45) = 1.49, p > .1). The second delay was significantly longer than the first (t(46) = 7.90, p < .001), resulting from the delay of MRI scanning because of hardware upgrades.

Data Conditioning

Participant age and the ICV-adjusted SUB, CA1-2, and CA3-4/DG volumes were centered at their respective sample means. In the absence of specific hypotheses regarding hemispheric differences in the associations between subfield volumes, memory, and age or the modification thereof by diagnosed hypertension, left and right hemisphere volumes were summed. Immediate and delayed subtest scores on WJ-R Memory for Names were summed to produce an index of the total performance on the measure (NamesTotal). All scores were standardized to minimize the potential influence of highly disparate scales.

Repeated Exposure Effects

Using Systat 13 (Systat Software, Inc., Chicago, IL), we computed residualized gain scores for memory tasks repeatedly presenting identical content over successive longitudinal measurements, including SDFR noun list tests and NamesImmediate. We calculated residualized gain scores by regressing longitudinal test scores from T3 on baseline scores from 4 years prior (T1) and extracting the residuals. One prior study (Woodard et al., 2010) used cutoffs in residualized gain scores of memory performance as the diagnostic of cognitive decline. In contrast, the inclusion of residualized gain scores in this study is intended to reflect individual differences in the retention of associative and verbal information over successive 2-year intervals. Moreover, such analysis should assist in the interpretation of longitudinal data likely contaminated by such effects.

Statistical Analyses

Parts of the data were missing in three cases that did not have immediate noun list recall scores for T1, one of which also did not have scores for delayed recall at T3 and two other cases lacked those scores as well.

The significant bivariate associations in normotensive and hypertensive subsamples are presented in Table 2, along with their confidence intervals (CIs) that were generated by bias-corrected (BC) bootstrapping (5,000 draws) with pairwise deletion. Correlations were significant if 95% CIs did not include zero. Table 2 is included for descriptive purposes, as bivariate correlations neither account for the influences of the other variables modeled nor estimate error as well as structural equation modeling approaches. Moreover, the limited sample size precludes testing of dissociations or moderation using correlations in Table 2.

Table 2. 

Significant Correlations by Hypertension Status

Variable 1
Variable 2
Bootstrapped Pearson's r
95% CI
Hypertensives 
Age CA1-2 −.592 −0.773, −0.326 
Age Noun list: SDFR, repeated list −.439 −0.735, −0.040 
CA1-2 vol. Noun list: LDFR, nonrepeated list .450 0.028, 0.749 
CA1-2 vol. Noun list: SDFR gain score .567 0.133, 0.836 
CA3-4/DG vol. Memory for Names: total score .550 0.287, 0.750 
CA3-4/DG vol. Memory for Names: immed. gain score .531 0.183, 0.747 
EC vol. Noun list: SDFR, nonrepeated list .404 0.163, 0.618 
EC vol. Noun list: LDFR, nonrepeated list .483 0.133, 0.747 
 
Normotensives 
Age Noun list: SDFR, repeated list −.520 −0.715, −0.158 
Age Noun list: SDFR, nonrepeated list −.486 −0.674, −0.227 
Age Noun list: LDFR, nonrepeated list −.431 −0.676, −0.100 
Age Memory for Names: total score −.617 −0.801, −0.229 
CA1-2 vol. CA3-4/DG vol. .515 0.051, 0.751 
Variable 1
Variable 2
Bootstrapped Pearson's r
95% CI
Hypertensives 
Age CA1-2 −.592 −0.773, −0.326 
Age Noun list: SDFR, repeated list −.439 −0.735, −0.040 
CA1-2 vol. Noun list: LDFR, nonrepeated list .450 0.028, 0.749 
CA1-2 vol. Noun list: SDFR gain score .567 0.133, 0.836 
CA3-4/DG vol. Memory for Names: total score .550 0.287, 0.750 
CA3-4/DG vol. Memory for Names: immed. gain score .531 0.183, 0.747 
EC vol. Noun list: SDFR, nonrepeated list .404 0.163, 0.618 
EC vol. Noun list: LDFR, nonrepeated list .483 0.133, 0.747 
 
Normotensives 
Age Noun list: SDFR, repeated list −.520 −0.715, −0.158 
Age Noun list: SDFR, nonrepeated list −.486 −0.674, −0.227 
Age Noun list: LDFR, nonrepeated list −.431 −0.676, −0.100 
Age Memory for Names: total score −.617 −0.801, −0.229 
CA1-2 vol. CA3-4/DG vol. .515 0.051, 0.751 

All scores are from T3. Only associations including age or subfield volumes are presented; significant associations between memory indices are not shown. Abbreviations: vol = volume; immed = immediate.

To examine such relationships, we used a structural equation modeling approach implemented in Mplus 7.0 (Muthén & Muthén, 2012). We fit grouped path analytic models to the data to investigate the differential associations between age, hippocampal subfield volumes, and memory as a function of diagnosed hypertension. Rather than including diagnosis as a dummy-coded variable, grouped analysis initially specifies a model including multiple hypothesized paths reflecting heterogeneous partitioning of variance based on group membership. First, we performed grouped path analysis to investigate the roles of CA1-2 and CA3-4/DG subfield volumes as contributors to the associations of age and performance on different tests of episodic memory; SUB was not included in the path models because of lack of significant zero-order associations for either group. Nested models were produced for each group by constraining nonhypothesized paths to zero; such an approach allows groups to feature different patterns of associations among the variables nested under a common model.

Path models used BC bootstrap resampling with 5,000 draws to generate estimates and 95% CIs of indirect effects (MacKinnon & Fairchild, 2009; Williams & MacKinnon, 2008; Cheung & Lau, 2007); indirect effects are calculated as the product of two or more direct paths. Many consider BC bootstrapped indirect effects to provide superior estimates of mediation (MacKinnon & Fairchild, 2009; Williams & MacKinnon, 2008; MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002; Shrout & Bolger, 2002) over other common methods (e.g., Baron & Kenny, 1986; Sobel, 1982).

Path Model Specification

On the basis of previous findings and on the results of bootstrapped zero-order correlations, we specified a path model that included seven continuous variables: age at T3 (AgeT3); EC volume; CA1-2 volume (CA1-2); CA3-4/DG volume (CA3-4/DG); scores on Memory for Names, total correct at T3 (NamesTotal); number of nouns from the nonrepeated list recalled following delay (NounLDFR); and the residualized gain scores for memory for names immediate memory subtest (NamesGain) and SDFR on the repeated noun lists (SDFRGain). To test age effects, we specified paths from AgeT3 to CA1-2, NamesTotal, NamesGain, NounLDFR, and SDFRGain. To model the effects of subfield volumes on memory performance, we specified paths from CA1-2 to both NounLDFR and SDFRGain and paths from CA3-4/DG to both NamesTotal and NamesGain. Model specification also included correlational relationships between several pairs of variables: CA1-2–CA3-4/DG, AgeT3–CA3-4/DG, NounLDFR–CA3-4/DG, and NounLDFR–SDFRGain. Correlational paths were also specified between EC and the four memory variables and AgeT3. A path was also specified from EC to CA3-4/DG volume. For normotensive participants, we constrained the path from CA1-2 to NounLDFR to zero; in the model for hypertensive participants, zero-constrained paths included those from AgeT3 to both NounLDFR and SDFRGain and from CA3-4/DG to NounLDFR.

RESULTS

In the comparison of unadjusted means, normotensive and hypertensive subgroups differed neither in memory (Table 1) nor in volumes of hippocampal subfields. There was a nonsignificant trend toward smaller CA1-2 subfield volumes in hypertensive participants. Notably, among individuals who were treated for hypertension, CA1 volume correlated negatively with systolic blood pressure: r = −.467, 95% CI [−0.745, −0.108]. However, there was no association between duration of anti-hypertensive treatment and subfield volumes, as all 95% CIs overlapped with zero. To decompose the variance among multiple correlated variables, we conducted path analyses as described below.

Path Analyses

The hypothesized path model testing different variance/covariance structures of age-subfield–memory associations by diagnosis of hypertension (Figure 2) fits the data very well, according to all goodness-of-fit indices (Table 3). Neither correlational nor regression paths between EC and any other variable were significant. Thus, to preserve degrees of freedom, to establish a more parsimonious model, and to maintain theoretical consistency with previous work that showed no cross-sectional associations between EC volume and memory in healthy aging (Rodrigue & Raz, 2004), we removed EC from the model. The model was re-fit including all other relations described above. The reduced model retained an excellent fit to the data. In comparison, the reversed-paths model that postulated the effect of cognitive variables on the brain showed a marginally acceptable fit, whereas the correlational model did not fit the data by any standard. Furthermore, comparison of all indices shows that the hypothesized model fits the data better than all other tested models.

Figure 2. 

Significant effects in path models grouped by hypertension diagnosis. Regression path values are standardized coefficients. Path models for normotensive (top) and hypertensive (bottom) participants. All variables represent scores from the third longitudinal wave of testing (T3), except gain scores that were calculated by regressing longitudinal test scores from T3 on baseline scores and extracting the residuals. ***p < .001, **p < .01, *p < .05.

Figure 2. 

Significant effects in path models grouped by hypertension diagnosis. Regression path values are standardized coefficients. Path models for normotensive (top) and hypertensive (bottom) participants. All variables represent scores from the third longitudinal wave of testing (T3), except gain scores that were calculated by regressing longitudinal test scores from T3 on baseline scores and extracting the residuals. ***p < .001, **p < .01, *p < .05.

Table 3. 

Goodness-of-Fit Indices

Model
χ2
df
χ2/df
p
CFI
TLI
RMSEA
SRMR
Path Models with Gain Scores 
Hypothesized model with EC 4.36 13 0.34 .987 1.00 1.40 .000 .044 
Reduced hypothesized model 5.07 0.56 .828 1.00 1.19 .000 .043 
Reversed model 9.72 1.08 .374 0.99 0.96 .057 .100 
Correlational model 48.88 20 2.44 .000 0.68 0.33 .243 .224 
Model
χ2
df
χ2/df
p
CFI
TLI
RMSEA
SRMR
Path Models with Gain Scores 
Hypothesized model with EC 4.36 13 0.34 .987 1.00 1.40 .000 .044 
Reduced hypothesized model 5.07 0.56 .828 1.00 1.19 .000 .043 
Reversed model 9.72 1.08 .374 0.99 0.96 .057 .100 
Correlational model 48.88 20 2.44 .000 0.68 0.33 .243 .224 

CFI = Comparative Fit Index; TLI = Tucker–Lewis Index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

The hypothesized model contained a significant positive path from CA3-4/DG volume to gain in names recall over repeated administrations (NamesGain), regardless of the hypertension diagnosis. For normotensive participants, age predicted a significant proportion of variance in the total memory for names (NamesTotal) score; in contrast, individual differences in NamesTotal among the hypertensive participants were associated with CA3-4/DG volume.

In the hypertensive group, there were significant indirect effects of age on noun recall and practice gains in short-term noun recall via CA1-2 volume: AgeT3 → CA1-2 volume → NounLDFR (std. estimate = −.267, p = .034, 95% CI [−0.513, −0.021]) and AgeT3 → CA1-2 Volume → SDFRGain (std. estimate = −.362, p = .004, 95% CI [−0.605, −0.119]). Older hypertensive participants with smaller CA1-2 volumes evidenced poorer long-delay noun recall and lower gain scores on repeated short-delay recall. In contrast, among the normotensives, the residualized gain scores for short-delay noun recall were not significantly associated with any other variable in the model.

DISCUSSION

The main finding in this study is the dissociation between neuroanatomical correlates of two memory tasks—free recall and associative recognition—in middle-aged and older adults. Memory for arbitrary picture–name associations was linked to the volume of the hippocampal area that encompasses an established adult neurogenesis region, that is, the DG. In contrast, free recall of common nouns was unrelated to the DG–CA3/4 volume and linked to the volume of another region, CA1-2, but only in persons with a common vascular risk factor.

The observed link between CA3-4/DG volumes and associative memory is in accord with the current view of that region's role in memory formation, whether it is conceptualized as a management of pattern separation and pattern completion processes (Yassa & Stark, 2011) or as a division of labor in handling low- versus high-resolution memories (Aimone, Deng, & Gage, 2011). Consistent with these models, it is plausible that the formation of arbitrary associations between nonsense “names” and novel cartoon characters that are required by the names test used in our study would rely primarily on the DG–CA3 system. According to a current conceptualization, that system is crucial for encoding novel stimuli and handles initial low-resolution memories (Aimone et al., 2011). Thus, our finding may reflect the fact that DG, a structure uniquely capable of producing new processing units by neurogenesis, may be critical for success in an associative memory task but is of little relevance to the free recall of familiar word stimuli. Individuals with larger DG (and CA3-4) may be better positioned to benefit from repeated presentations of arbitrary associations than their peers with smaller volumes of those structures. The lack of correlation between gain scores and age was in accord with previous reports (Salthouse, 2013) and suggested that CA3-4/DG may play a role in signal maintenance or enhancement of previously learned associative representations over very long delays. In previous studies, larger CA3-4/DG volumes predicted better discriminability and reduced false alarm rate associative recognition (Mueller et al., 2011; Shing et al., 2011). Notably, whereas both previous studies (Mueller et al., 2011; Shing et al., 2011) used only verbal material, that is, familiar stimuli forming unfamiliar associations, our findings generalize the relationship between memory and CA3-4/DG volume to novel verbal and pictorial stimuli.

In contrast to the observed link between CA3-4/DG and associative memory, the association between CA1-2 volume and performance on noun recall task was noted only in the group of participants with a history of hypertension. This finding is likely to reflect this region's vulnerability to vascular risk (Schmidt-Kastner & Freund, 1991). Although association between free recall performance and CA3-4/DG volume has been reported in one study (Mueller et al., 2011), it was not with respect to the number of correctly recalled items but the discriminability and false alarm rate, that is, indices related to memory resolution and pattern separation.

On the basis of cross-sectional evidence, Mueller and colleagues (2011) suggested that the differential associations between hippocampal regions and memory reflected distinct temporal characteristics of memory processes, with CA3-4/DG supporting delayed recent representations and CA1-2 contributing to consolidation and delayed recall. The present findings based on the examination of long-term changes in memory performance provide no support for that hypothesis. Performance at the most recent measurement occasion and long-term gains in associative memory were linked to CA3-4/DG volumes, whereas a similar pattern of associations was observed for the free recall of nouns and CA1-2 volumes, although only among hypertensive participants. Among normotensive participants, only gains in associative memory were related to a neuroanatomical measure, CA3-4/DG volume. In short, neuroanatomical correlates of memory depended on the nature of the test and participant characteristics rather than the temporal characteristic of testing. Given the nature of the memory task and imaging and volumetric methods, disambiguation of pattern completion and pattern separation processes and specific contributions of CA3, CA4, and DG were not feasible. Increased false alarm rates could result from faulty pattern completion processes. Associative false recognition could signify the incorrectly strong signal that items were associated with or may be related to an increased reliance on familiarity (Addante, Ranganath, Olichney, & Yonelinas, 2012; Yonelinas et al., 2002).

Our findings reinforce the need for incorporating vascular risk factors in the investigations of age differences in the brain and the role of the brain in age-related differences in memory. This may be especially relevant when the study sample includes persons with declining cognitive performance. For example, the extant reports of smaller CA1-2 volumes in older adults (Mueller & Weiner, 2009) may reflect inclusion of individuals with risk factors for further decline. At least one previous study found that age differences in CA1-2 volumes were entirely attributable to the presence of hypertensive individuals in the older group (Shing et al., 2011). The observed association between systolic blood pressure and CA1-2 volumes suggests that the effect may be dose dependent, that is, persons with poorer control of hypertension are more likely to suffer damage to CA1. However, in the absence of longitudinal follow-up, this conjecture is speculative.

Essential hypertension is the strongest predictor of stroke and an antecedent of cerebrovascular disease (Staessen, Kuznetsova, & Stolarz, 2003; Wolf-Maier et al., 2003), which is in turn linked to reduced cerebral blood volume in CA1 (Wu et al., 2008). Hypertension is associated with reductions in CBF, which may be improved by medical treatment (Efimova, Efimova, Triss, & Lishmanov, 2008; Beason-Held, Moghekar, Zonderman, Kraut, & Resnick, 2007; Korf, White, Scheltens, & Launer, 2004). Negative associations between memory and hippocampal integrity in the presence of vascular pathology are frequently reported (Small, Schobel, Buxton, Witter, & Barnes, 2011). Notably, patients with hippocampal atrophy resulting from hypoxia–ischemia after cardiac arrest exhibit greater deficits in free recall than in recognition (Yonelinas et al., 2002) and evidence source memory deficits with no impairment in familiarity (Addante et al., 2012). Therefore, increased vascular risk may confer particular neuroanatomical vulnerability on selected sectors of the hippocampus and differentially affects a specific type of episodic memory—free recall. Persons with larger DG and CA3 sectors benefited more from repeated exposure, whereas only age was related to the concurrent level of performance in normotensive individuals. Moreover, DG granule cell neurogenesis is significantly reduced in rat models of hypertension (Pietranera, Lima, Roig, & De Nicola, 2010), although it is unclear how this would be affected by treatment or duration of condition in humans. In this study, only among participants treated for hypertension, CA3-4/DG volume predicted concurrent associative memory performance, even after controlling for practice effects. Thus, although the history of hypertension did not modify the relationship between CA3-4/DG volume and longitudinal improvement in associative memory, the relationship between concurrent associative memory and CA3-4/DG volume depended on its presence.

Limitations and Future Directions

Although a significant improvement over the whole hippocampus measures, the present methods are still crude and limited. For example, our rules of regional demarcation treat CA3, CA4, and DG as one unit and do not permit testing of hypotheses regarding the differential roles of each subfield in pattern separation and pattern completion. Further refinement of image acquisition and measurement techniques may help to resolve this issue.

Several contributors to individual differences in memory and hippocampal anatomy were left unexamined in this study. For example, we did not account for ApoE genotype, which may explain a significant proportion of variance in regional hippocampal volumes and memory by itself (Mueller & Weiner, 2009) or in interaction with elevated vascular risk (Bender & Raz, 2012a, 2012b). In light of the power limitations in this study, we restricted the number of variables and estimated parameters to test specific hypotheses regarding associations between CA1-2 and CA3-4/DG subfield volumes and memory. Other studies have focused more on the dissociations between volumes of subfields and adjacent cortical areas in memory, pathology, and aging (Mueller, Schuff, Raptentsetsang, Elman, & Weiner, 2008). Because the hippocampus is only one part of a complex system, future efforts aimed at understanding associations of brain structure with memory and their modification by vascular risk should also account for such effects on adjacent medial-temporal lobe structures and other regions. In addition, the imaging method used in this study, while allowing for a high in-plane resolution, acquired thick sections of the brain and did not cover the whole hippocampus. Thus, we could not compare the associations between volume and memory for the regions and hippocampus in toto. At the time of this writing, these measures are unavailable, although they will become available in the future and will allow us to revisit this question.

The administration of the noun-list free recall task changed from the first administration to subsequent waves of testing. At baseline, only short-term recall of nouns was tested, whereas the second and third assessments included both SDFR and LDFR tasks. Future studies should assess change in both short- and long-delayed free recall abilities using both repeated and novel stimuli at each assessment.

The sample for this study was selected for optimal self-reported health history. Thus, the findings may not generalize to the larger population in which advanced age is frequently accompanied by metabolic and cardiovascular disease. Similarly, the limited sample size and large proportion of female participants in this study may also limit the generalizability of these findings. Moreover, this study that uses longitudinal memory assessments is likely biased by self-selection and selective attrition toward higher performing healthier participants (Baltes, 1968).

Hypertension was the sole vascular risk factor examined here, and subsequent studies should explore the modification of longitudinal change by individual differences in additional vascular risk markers. Our small sample size precluded a detailed examination of the effects of specific antihypertensive medications on memory and hippocampal volumes. The impact of medical treatment of hypertension on neuroanatomical and cognitive variables is unclear. According to some studies, pharmaceuticals may blunt the impact of hypertension on hippocampal atrophy (Korf et al., 2004). However, a recent small-scale control trial yielded a less optimistic conclusion (Jennings et al., 2012), suggesting that, by the time treatment is initiated, long-standing hypertension has already inflicted irreversible damage. Moreover, because it is possible that individual differences in hypertension history and medication response may account for variance in regional hippocampal volumes and cognition, future studies should evaluate interactions between change in volume and continuous measures of vascular risk in participants not taking antihypertensive drugs.

In summary, in a sample of healthy middle-aged and older adults, we found that participants with larger CA3-4/DG volumes benefited more from repeated testing of associative memory, regardless of the vascular risk and age. In the presence of elevated vascular risk, age differences in CA1-2 volume predicted variance in delayed free recall. The results underscore the need for finer measurements of brain structure and attention to vascular risk factors in studies of neural mechanisms of cognitive aging.

Acknowledgments

This work was supported in part by a grant from the National Institute on Aging R37-AG011230 to N. R. We thank Cheryl Dahle for assistance in collection of the cognitive data.

Reprint requests should be sent to Naftali Raz, 87 E. Ferry Street, 226 Knapp Building, Detroit, MI 48202, or via e-mail: nraz@wayne.edu.

REFERENCES

Addante
,
R. J.
,
Ranganath
,
C.
,
Olichney
,
J.
, &
Yonelinas
,
A. P.
(
2012
).
Neurophysiological evidence for a recollection impairment in amnesia patients that leaves familiarity intact.
Neuropsychologia
,
50
,
3004
3014
.
Aimone
,
J. B.
,
Deng
,
W.
, &
Gage
,
F. H.
(
2011
).
Resolving new memories: A critical look at the dentate gyrus, adult neurogenesis, and pattern separation.
Neuron
,
70
,
589
596
.
Baltes
,
P. B.
(
1968
).
Longitudinal and cross-sectional sequences in the study of age and generation effects.
Human Development
,
11
,
145
171
.
Baron
,
R. M.
, &
Kenny
,
D. A.
(
1986
).
The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations.
Journal of Personality and Social Psychology
,
51
,
1173
1182
.
Beason-Held
,
L. L.
,
Moghekar
,
A.
,
Zonderman
,
A. B.
,
Kraut
,
M. A.
, &
Resnick
,
S. M.
(
2007
).
Longitudinal changes in cerebral blood flow in the older hypertensive brain.
Stroke
,
38
,
1766
1773
.
Bender
,
A. R.
, &
Raz
,
N.
(
2012a
).
Age-related differences in episodic memory: A synergistic contribution of genetic and physiological vascular risk factors.
Neuropsychology
,
26
,
442
.
Bender
,
A. R.
, &
Raz
,
N.
(
2012b
).
Age-related differences in memory and executive functions in healthy APOE varepsilon4 carriers: The contribution of individual differences in prefrontal volumes and systolic blood pressure.
Neuropsychologia
,
50
,
704
714
.
Birns
,
J.
, &
Kaira
,
L.
(
2008
).
Cognitive function and hypertension.
Journal of Human Hypertension
,
23
,
86
96
.
Brady
,
C. B.
,
Spiro
,
A.
, III, &
Gaziano
,
J. M.
(
2005
).
Effects of age and hypertension status on cognition: The Veterans Affairs normative aging study.
Neuropsychology
,
19
,
770
777
.
Cheung
,
G. W.
, &
Lau
,
R. S.
(
2007
).
Testing mediation and suppression effects of latent variables: Bootstrapping with structural equation models.
Organizational Research Methods
,
11
,
296
325
.
Chobanian
,
A. V.
,
Bakris
,
G. L.
,
Black
,
H. R.
,
Cushman
,
W. C.
,
Green
,
L. A.
,
Izzo
,
J. L.
,
et al
(
2003
).
Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure.
Hypertension
,
42
,
1206
1252
.
Efimova
,
I. Y.
,
Efimova
,
N. Y.
,
Triss
,
S. V.
, &
Lishmanov
,
Y. B.
(
2008
).
Brain perfusion and cognitive function changes in hypertensive patients.
Hypertension Research
,
31
,
673
678
.
Elias
,
M. F.
,
Robbins
,
M. A.
,
Elias
,
P. K.
, &
Streeten
,
D. H.
(
1998
).
A longitudinal study of blood pressure in relation to performance on the Wechsler Adult Intelligence Scale.
Health Psychology
,
17
,
486
493
.
Folstein
,
M. F.
,
Folstein
,
S. E.
, &
McHugh
,
P. R.
(
1975
).
“Mini-mental state.” A practical method for grading the cognitive state of patients for the clinician.
Journal of Psychiatric Research
,
12
,
189
198
.
Friendly
,
M.
,
Franklin
,
P. E.
,
Hoffman
,
D.
, &
Rubin
,
D. C.
(
1982
).
The Toronto Word Pool: Norms for imagery, concreteness, orthographic variables, and grammatical usage for 1,080 words.
Behavior Research Methods & Instrumentation
,
14
,
375
399
.
Hannesdottir
,
K.
,
Nitkunan
,
A.
,
Charlton
,
R. A.
,
Barrick
,
T. R.
,
MacGregor
,
G. A.
, &
Markus
,
H. S.
(
2009
).
Cognitive impairment and white matter damage in hypertension: A pilot study.
Acta Neurologica Scandinavica
,
119
,
261
268
.
Harrington
,
F.
,
Saxby
,
B. K.
,
McKeith
,
I. G.
,
Wesnes
,
K.
, &
Ford
,
G. A.
(
2000
).
Cognitive performance in hypertensive and normotensive older subjects.
Hypertension
,
36
,
1079
1082
.
Jenkinson
,
M.
,
Pechaud
,
M.
, &
Smith
,
S.
(
2005, June
).
BET2: MR-based estimation of brain, skull and scalp surfaces.
In
Eleventh annual meeting of the Organization for Human Brain Mapping
(
Vol. 17
).
Jennings
,
J. R.
,
Mendelson
,
D. N.
,
Muldoon
,
M. F.
,
Ryan
,
C. M.
,
Gianaros
,
P. J.
,
Raz
,
N.
,
et al
(
2012
).
Regional grey matter shrinks in hypertensive individuals despite successful lowering of blood pressure.
Journal of Human Hypertension
,
26
,
295
305
.
Keihaninejad
,
S.
,
Heckemann
,
R. A.
,
Fagiolo
,
G.
,
Symms
,
M. R.
,
Hajnal
,
J. V.
, &
Hammers
,
A.
(
2010
).
A robust method to estimate the intracranial volume across MRI field strengths (1.5 T and 3 T).
Neuroimage
,
50
,
1427
1437
.
Kirino
,
T.
, &
Sirno
,
K.
(
1984
).
Selective vulnerability in the gerbil hippocampus following transient ischmeia.
Acta Neuropathologica
,
62
,
201
208
.
Knopman
,
D.
,
Boland
,
L. L.
,
Mosley
,
T.
,
Howard
,
G.
,
Liao
,
D.
,
Szklo
,
M.
,
et al
(
2001
).
Cardiovascular risk factors and cognitive decline in middle-aged adults.
Neurology
,
56
,
42
48
.
Korf
,
E. S. C.
,
White
,
L. R.
,
Scheltens
,
P.
, &
Launer
,
L. J.
(
2004
).
Midlife blood pressure and the risk of hippocampal atrophy the Honolulu Asia aging study.
Hypertension
,
44
,
29
34
.
Kuo
,
H. K.
,
Sorond
,
F.
,
Iloputaife
,
I.
,
Gagnon
,
M.
,
Milberg
,
W.
, &
Lipsitz
,
L. A.
(
2004
).
Effect of blood pressure on cognitive functions in elderly persons.
The Journals of Gerontology, Series A, Medical Sciences
,
59
,
1191
1194
.
MacKinnon
,
D. P.
, &
Fairchild
,
A. J.
(
2009
).
Current directions in mediation analysis.
Current Directions in Psychological Science
,
18
,
16
20
.
MacKinnon
,
D. P.
,
Lockwood
,
C. M.
,
Hoffman
,
J. M.
,
West
,
S. G.
, &
Sheets
,
V.
(
2002
).
A comparison of methods to test mediation and other intervening variable effects.
Psychological Methods
,
7
,
83
104
.
Moscovitch
,
M.
,
Nadel
,
L.
,
Winocur
,
G.
,
Gilboa
,
A.
, &
Rosenbaum
,
R. S.
(
2006
).
The cognitive neuroscience of remote episodic, semantic and spatial memory.
Current Opinion in Neurobiology
,
16
,
179
190
.
Mueller
,
S. G.
,
Chao
,
L. L.
,
Berman
,
B.
, &
Weiner
,
M. W.
(
2011
).
Evidence for functional specialization of hippocampal subfields detected by MR subfield volumetry on high resolution images at 4 T.
Neuroimage
,
56
,
851
857
.
Mueller
,
S. G.
,
Schuff
,
N.
,
Raptentsetsang
,
S.
,
Elman
,
J.
, &
Weiner
,
M. W.
(
2008
).
Selective effect of Apo e4 on CA3 and dentate in normal aging and Alzheimer's disease using high resolution MRI at 4 T.
Neuroimage
,
42
,
42
48
.
Mueller
,
S. G.
,
Stables
,
L.
,
Du
,
A. T.
,
Schuff
,
N.
,
Truran
,
D.
,
Cashdollar
,
N.
,
et al
(
2007
).
Measurement of hippocampal subfields and age-related changes with high resolution MRI at 4 T.
Neurobiology of Aging
,
28
,
719
726
.
Mueller
,
S. G.
, &
Weiner
,
M. W.
(
2009
).
Selective effect of age, Apo e4, and Alzheimer's disease on hippocampal subfields.
Hippocampus
,
19
,
558
564
.
Muthén
,
L.
, &
Muthén
,
B.
(
2012
).
Mplus user's guide
(6th ed.).
Los Angeles, CA
:
Muthén & Muthén
.
Nitatori
,
T.
,
Sato
,
N.
,
Waguri
,
S.
,
Karasawa
,
Y.
,
Araki
,
H.
,
Shinbani
,
K.
,
et al
(
1995
).
Delayed neuronal death in the CA1 pyramidal cell layer of the gerbil hippocampus following transient ischemia is apoptosis.
The Journal of Neuroscience
,
15
,
1001
1011
.
Old
,
S. R.
, &
Naveh-Benjamin
,
M.
(
2008
).
Differential effects of age on item and associative measures of memory: A meta-analysis.
Psychology and Aging
,
23
,
104
118
.
Oldfield
,
R. C.
(
1971
).
The assessment and analysis of handedness: The Edinburgh inventory.
Neuropsychologia
,
9
,
97
113
.
Pickering
,
T. G.
,
Hall
,
J. E.
,
Appel
,
L. J.
,
Falkner
,
B. E.
,
Graves
,
J.
,
Hill
,
M. N.
,
et al
(
2005
).
Recommendations for blood pressure measurement in humans and experimental animals.
Hypertension
,
45
,
142
161
.
Pietranera
,
L.
,
Lima
,
A.
,
Roig
,
P.
, &
De Nicola
,
A. F.
(
2010
).
Involvement of brain-derived neurotrophic factor and neurogenesis in oestradiol neuroprotection of the hippocampus of hypertensive rats.
Journal of Neuroendocrinology
,
22
,
1082
1092
.
Radloff
,
L. S.
(
1977
).
The CES-D scale: A self-report depression scale for research in the general population.
Applied Psychological Measurement
,
1
,
385
401
.
Raz
,
N.
, &
Kennedy
,
K. M.
(
2009
).
A systems approach to the aging brain: Neuroanatomic changes, their modifiers, and cognitive correlates.
In W. Jagust & M. D'Esposito (Eds.)
,
Imaging the aging brain
(pp.
43
70
).
New York
:
Oxford University Press
.
Raz
,
N.
,
Lindenberger
,
U.
,
Rodrigue
,
K. M.
,
Kennedy
,
K. M.
,
Head
,
D.
,
Williamson
,
A.
,
et al
(
2005
).
Regional brain changes in aging healthy adults: General trends, individual differences and modifiers.
Cerebral Cortex
,
15
,
1676
1689
.
Raz
,
N.
, &
Rodrigue
,
K.
(
2006
).
Differential aging of the brain: Patterns, cognitive correlates and modifiers.
Neuroscience & Biobehavioral Reviews
,
30
,
730
748
.
Raz
,
N.
,
Rodrigue
,
K. M.
, &
Acker
,
J. D.
(
2003
).
Hypertension and the brain: Vulnerability of the prefrontal regions and executive functions.
Behavioral Neuroscience
,
17
,
1169
1180
.
Rodrigue
,
K. M.
, &
Raz
,
N.
(
2004
).
Shrinkage of the entorhinal cortex over five years predicts memory performance in healthy adults.
Journal of Neuroscience
,
24
,
956
963
.
Sabbatini
,
M.
,
Strocchi
,
P.
,
Vitaioli
,
L.
, &
Amenta
,
F.
(
2000
).
The hippocampus in spontaneously hypertensive rats: A quantitative microanatomical study.
Neuroscience
,
100
,
251
258
.
Salthouse
,
T. A.
(
2013
).
Effects of first occasion test experience on longitudinal cognitive change.
Developmental Psychology
.
[Epub ahead of print]
.
Saxby
,
B. K.
,
Harrington
,
F.
,
McKeith
,
I. G.
,
Wesnes
,
K.
, &
Ford
,
G. A.
(
2003
).
Effects of hypertension on attention, memory, and executive function in older adults.
Health Psychology
,
22
,
587
591
.
Schmidt-Kastner
,
R.
, &
Freund
,
T. F.
(
1991
).
Selective vulnerability of the hippocampus in brain ischemia.
Neuroscience
,
40
,
599
636
.
Semplicini
,
A.
,
Inverso
,
G.
,
Realdi
,
A.
,
Macchini
,
L.
,
Maraffon
,
M.
,
Puato
,
M.
,
et al
(
2010
).
Blood pressure control has distinct effects on executive function, attention, memory and markers of cerebrovascular damage.
Journal of Human Hypertension
,
25
,
80
87
.
Shing
,
Y. L.
,
Rodrigue
,
K. M.
,
Kennedy
,
K. M.
,
Fandakova
,
Y.
,
Bodammer
,
N.
,
Werkle-Bergner
,
M.
,
et al
(
2011
).
Hippocampal subfield volumes: Age, vascular risk, and correlation with associative memory.
Frontiers in Aging Neuroscience
,
3
,
2
.
Shrout
,
P. E.
, &
Bolger
,
N.
(
2002
).
Mediation in experimental and nonexperimental studies: New procedures and recommendations.
Psychological Methods
,
7
,
422
445
.
Shrout
,
P. E.
, &
Fleiss
,
J. L.
(
1979
).
Intraclass correlations: Uses in assessing rater reliability.
Psychological Bulletin
,
86
,
420
428
.
Singhmanoux
,
A.
, &
Marmot
,
M.
(
2005
).
High blood pressure was associated with cognitive function in middle-age in the Whitehall II study.
Journal of Clinical Epidemiology
,
58
,
1308
1315
.
Small
,
S. A.
,
Schobel
,
S. A.
,
Buxton
,
R. B.
,
Witter
,
M. P.
, &
Barnes
,
C. A.
(
2011
).
A pathophysiological framework of hippocampal dysfunction in ageing and disease.
Nature Reviews Neuroscience
,
12
,
585
601
.
Smith
,
S. M.
(
2002
).
Fast robust automated brain extraction.
Human Brain Mapping
,
17
,
143
155
.
Sobel
,
M. E.
(
1982
).
Asymptotic confidence intervals for indirect effects in structural equation models.
In S. Leinhardt (Ed.)
,
Sociological methodology
(
Vol. 13
, pp.
290
312
).
Washington, DC
:
American Sociological Association
.
Spencer
,
W. D.
, &
Raz
,
N.
(
1995
).
Differential effects of aging on memory for content and context: A meta-analysis.
Psychology and Aging
,
10
,
527
539
.
Squire
,
L. R.
, &
Zola
,
S. M.
(
1996
).
Structure and function of declarative and nondeclarative memory systems.
Proceedings of the National Academy of Sciences, U.S.A.
,
93
,
13515
13522
.
Staessen
,
J. A.
,
Kuznetsova
,
T.
, &
Stolarz
,
K.
(
2003
).
Hypertension prevalence and stroke mortality across populations.
The Journal of the American Medical Association
,
289
,
2420
2422
.
Sugawara
,
T.
,
Lewen
,
A.
,
Noshita
,
N.
,
Gasche
,
Y.
, &
Chan
,
P. H.
(
2002
).
Effects of global ischemia duration on neuronal, astroglial, oligodendroglial, and microglial reactions in the vulnerable hippocampal CA1 subregion in rats.
Journal of Neurotrauma
,
19
,
85
98
.
Suyama
,
K.
(
1992
).
Changes of neuronal transmission in the hippocampus after transient ischemia in spontaneously hypertensive rats and the protective effects of MK-801.
Stroke
,
23
,
260
266
.
Van Petten
,
C.
(
2004
).
Relationship between hippocampal volume and memory ability in healthy individuals across the lifespan: Review and meta-analysis.
Neuropsychologia
,
42
,
1394
1413
.
Waldstein
,
S. R.
,
Brown
,
J. R.
,
Maier
,
K. J.
, &
Katzel
,
L. I.
(
2005
).
Diagnosis of hypertension and high blood pressure levels negatively affect cognitive function in older adults.
Annals of Behavioral Medicine
,
29
,
174
180
.
Waldstein
,
S. R.
, &
Katzel
,
L. I.
(
2001
).
Hypertension and cognitive function.
In S. R. Waldstein & M. F. Elias (Eds.)
,
Neuropsychology of cardiovascular disease
(pp.
15
36
).
Mahwah, NJ
:
Erlbaum
.
Williams
,
J.
, &
MacKinnon
,
D. P.
(
2008
).
Resampling and distribution of the product methods for testing indirect effects in complex models.
Structural Equation Modeling
,
15
,
23
51
.
Wills
,
A. K.
,
Lawlor
,
D. A.
,
Matthews
,
F. E.
,
Aihie Sayer
,
A.
,
Bakra
,
E.
,
Ben-Shlomo
,
Y.
,
et al
(
2011
).
Life course trajectories of systolic blood pressure using longitudinal data from eight UK cohorts.
PLoS Medicine
,
8
,
e1000440
.
Witter
,
M.
, &
Amaral
,
D.
(
1995
).
Hippocampal formation.
In G. Paxinos (Ed.)
,
The rat nervous system
(2nd ed., pp.
443
493
).
San Diego, CA
:
Academic Press
.
Wolf-Maier
,
K.
,
Cooper
,
R. S.
,
Banegas
,
J. R.
,
Giampaoli
,
S.
,
Hense
,
H. W.
,
Joffres
,
M.
,
et al
(
2003
).
Hypertension prevalence and blood pressure levels in 6 European countries, Canada, and the United States.
The Journal of the American Medical Association
,
289
,
2363
2369
.
Woodard
,
J. L.
,
Seidenberg
,
M.
,
Nielson
,
K. A.
,
Smith
,
J. C.
,
Antuono
,
P.
,
Durgerian
,
S.
,
et al
(
2010
).
Prediction of cognitive decline in healthy older adults using fMRI.
Journal of Alzheimer's Disease
,
21
,
871
885
.
Woodcock
,
R. W.
, &
Johnson
,
M. B.
(
1989
).
Woodcock–Johnson tests of cognitive ability.
Allen, TX
:
DLM Teaching Resources
.
Wu
,
W.
,
Brickman
,
A. M.
,
Luchsinger
,
J.
,
Ferrazzano
,
P.
,
Pichiule
,
P.
,
Yoshita
,
M.
,
et al
(
2008
).
The brain in the age of old: The hippocampal formation is targeted differentially by diseases of late life.
Annals of Neurology
,
64
,
698
706
.
Yassa
,
M. A.
, &
Stark
,
C. E.
(
2011
).
Pattern separation in the hippocampus.
Trends in Neurosciences
,
34
,
515
525
.
Yonelinas
,
A. P.
,
Kroll
,
N. E. A.
,
Quamme
,
J. R.
,
Lazzara
,
M. M.
,
Sauvé
,
M.-J.
,
Widaman
,
K. F.
,
et al
(
2002
).
Effects of extensive temporal lobe damage or mild hypoxia on recollection and familiarity.
Nature Neuroscience
,
5
,
1236
1241
.