The effect of hippocampal subfield volumes on cognitive decline and incident dementia in a memory clinic cohort

Abstract

Background

The hippocampus plays a central role in cognition and hippocampal atrophy is a key hallmark of Alzheimer's disease. Evidence has suggested associations between hippocampal subfield volumes and specific cognitive domains and dementia risk. However, to our knowledge, no study has examined the role of hippocampal subfield volumes in cognitive decline across different domains over time.

Objective

We investigated associations between hippocampal subfield volumes and changes in cognitive domains together with incident dementia in a memory clinic cohort.

Methods

Associations between hippocampal subfield volumes and cognitive decline over three years (n = 443) were analyzed using generalized estimating equations, and associations with incident dementia (n = 283) using multiple logistic regression.

Results

At baseline, all hippocampal subfield volumes were associated with diagnosis of dementia, while the CA4-dentate gyrus, molecular layer, subicular complex, and fimbria volumes were associated with diagnosis of CIND. Over three years, all subfields except the hippocampal fissure were associated with memory. Decreased molecular layer (OR:2.26, 95%CI:1.50;3.50) size was associated with increased risk of dementia.

Conclusions

Our findings suggest that hippocampal atrophy of the cornu ammonis, CA4-dentate gyrus, and molecular layer may first manifest with cognitive impairment in memory before other subfields of the hippocampus, and that molecular layer volume may be an early biomarker of dementia. Further research demonstrating the biological role of hippocampal subfields in specific cognitive domains is required.

Keywords

Alzheimer's disease cognitive decline dementia hippocampus magnetic resonance image

Introduction

The hippocampus plays a central role in cognition^1,2 and hippocampal atrophy is a key hallmark of Alzheimer's disease (AD).^3,4 Existing evidence shows that the accumulation of AD-related pathology such as amyloid deposition and neurofibrillary tangles has been associated with hippocampal atrophy.^5,6 Furthermore, smaller hippocampal volume has been associated with the risk of dementia, as well as with disease progression in dementia.⁷

Recently, accumulating evidence suggests that the hippocampus has distinct subfields associated with specialized cognitive functions, and that atrophy in different subfields were associated with different risks for dementia.⁸ Existing cross-sectional studies have shown that subfield volumes of the hippocampus such as the cornu ammonis area 1 (CA1) and CA2 were associated with memory,^9,10 the CA4 was associated with visuospatial memory,¹¹ the dentate gyrus was associated with memory and spatial navigation,¹² while the presubiculum and subiculum were associated with the pegboard test and the Stroop naming task.⁸ Furthermore, in the population-based Rotterdam Study, smaller volumes of the subiculum were specifically associated with incident dementia.⁸

Associations of the cornu ammonis, dentate gyrus, and subicular complex with memory decline and AD have been explained by the involvement of these subfields in the perforant and alvear pathways, connectional routes from the entorhinal cortex that are essential for the encoding and consolidation of memories.^13–17 The perforant pathway, through which inputs from the entorhinal cortex perforate the subicular complex to reach the dentate gyrus,¹⁴ has been found to be important for visual memory encoding and spatial information processing.^18,19 Associations between the perforant pathway and AD have been attributed to both structural changes in the dentate gyrus and the reduction of perforant pathway projections as a consequence of tau pathology in the entorhinal cortex.¹⁵ Evidence has also suggested that the alvear pathway, through which inputs from the entorhinal cortex project to the cornu ammonis,¹⁶ is important for memory consolidation,^17,20 and that pathological changes in this pathway are implicated in AD.²¹ The functions of the hippocampal subfields may thus explain differences in their associations with cognition.⁹

Apart from cross-sectional studies on the association between hippocampal subfield volumes and cognition, existing studies have evaluated the role of hippocampal subfields in incident dementia. However, to our knowledge, no study has examined the role of hippocampal subfield volumes in cognitive decline across different domains over time. While recent research has demonstrated the involvement of the hippocampus in various areas of cognition not limited to memory, further research is required to understand the effect of hippocampal subfield volumes on decline in other cognitive domains.^22,23 Existing data suggests that differential trajectories for cognitive decline in specific cognitive domains occur in both AD as well as vascular dementia.²⁴ Furthermore, the hippocampus may atrophy in a sequential pattern over time,²⁵ which supports the hypothesis that different subfields of the hippocampus may contribute to trajectories of cognitive decline in different cognitive domains over time.

Understanding the effect of hippocampal subfield volumes on changes in specific cognitive domains can provide new insights into how different aspects of cognition are supported by specific subfields within the hippocampus. This can also offer pathophysiological insights into the role of hippocampal subfield atrophy on cognitive impairment and dementia, shed light on individual variability to determine whether certain subfields are more susceptible to age-related alterations than others, and aid in developing targeted interventions in the future. Thus, we aimed to investigate the associations between baseline hippocampal subfield volumes with changes in specific cognitive domains as well as the risk of dementia over a three-year follow up in a prospective memory clinic cohort. Based on the existing international literature, we hypothesized that smaller baseline subicular complex, cornu ammonis and CA4-dentate gyrus volumes would be associated with decline in memory^9–11,26,27 and that smaller baseline CA4-dentate gyrus volume would be associated with decline in visuospatial function.^12,28 In addition, we hypothesized that baseline subicular complex volume would be associated with incident dementia.^8,29

Methods

Study population and cognitive diagnoses

Patients were recruited from an ongoing memory clinic at the National University Hospital, Singapore. Participants included in the study had three diagnostic categories at baseline assessment³⁰: (i) no cognitive impairment (NCI) defined as patients with no functional loss or cognitive impairment on neuropsychological tests; (ii) cognitive impairment with no dementia (CIND) defined as patients with impairment in at least one cognitive domain and who did not have loss of daily functions; and (iii) dementia defined according to the Diagnostic and Statistical Manual of Mental Disorders–4th Edition criteria. Participants were considered to have impairment in a cognitive domain if they scored 1.5 standard deviations (SD) below education-adjusted cut-off values in at least half of the individual tests in that domain.³¹ Cognitive diagnoses were assessed again for all patients after two years of follow-up. Patients with progression from NCI or CIND to dementia were defined as incident dementia.

A total of 573 subjects were recruited into the memory clinic cohort from August 2010 to October 2015. Of these, 39 (6.8%) patients were excluded due to missing data on hippocampal subfields, 92 (16.1%) patients were excluded due to missing data on follow-up cognitive diagnoses, and 91 (15.9%) were excluded due to missing data on cognitive domain scores at two or more follow-up visits. A total of 534 (93.2%) patients were included in the cross-sectional analysis of hippocampal subfields with cognitive domains, 443 (77.3%) patients were included in the longitudinal analysis of hippocampal subfields with cognitive domains, and 283 (49.4%) patients were included in the analysis of incident dementia (Figure 1). Institutional ethics approval (National Healthcare Group Domain Specific Review Board) was obtained prior to the conduct of the study, and written informed consent was obtained from all patients. Consent for patients with dementia was provided by their legal representative.

Figure 1.

Flow chart on patient selection from the memory clinic cohort and cases included in the analyses.

Demographic and cardiovascular risk assessment

All patients underwent a standardized clinical and neuropsychological examination at baseline. Data were collected using a questionnaire that included information on age, gender, race, and education. Past medical history of cardiovascular disease (defined as patients who had a history of ischemic heart disease, congestive heart failure, or atrial fibrillation) was recorded and subsequently verified via the patient's medical records.

Neuroimaging protocol and MRI biomarkers of cerebrovascular disease

MRI was performed at baseline using a 3-T Siemens Magnetom Trio Tim scanner with a 32-channel head coil at the Clinical Imaging Research Centre, National University of Singapore. The standardized neuroimaging protocol included three-dimensional (3D) T1-weighted imaging, T2-weighted imaging, susceptibility-weighted imaging (SWI), and fluid-attenuated inversion recovery (FLAIR) imaging.

MRI markers of cerebrovascular disease were graded based on the Standards for Reporting Vascular Changes on Neuroimaging (STRIVE) criteria and included lacunar infarcts, cortical infarcts, and white matter hyperintensities.³² The detailed image acquisition protocols for the quantification of these markers have been described previously.^33,34 The weighted T1 sequence and T2-weighted images were utilized to quantify brain tissue segmentation, while WMH volume was segmented using the FLAIR sequence.^35–38

Quantification of hippocampal subfield volumes

Hippocampal volume was derived using a model-based automated procedure (FreeSurfer v.6.0) on T1-weighted MRI.³⁹ Using an ultra-high resolution ex-vivo atlas, the hippocampus was segmented into eight subfields, namely the cornu ammonis (CA1 and CA2-3),⁴⁰ CA4-dentate gyrus (CA4 and the molecular and granule cell layers of the dentate gyrus),^41,42 hippocampal-amygdala transition area (HATA), molecular layer, subicular complex (subiculum, presubiculum, and parasubiculum),⁴³ fissure, fimbria, and tail (Figure 2).

Figure 2.

Hippocampal subfield segmentation. Note: Illustration of the hippocampal subfield segmentation using FreeSurfer v.6.0 in (A) sagittal, (B) axial, and (C) coronal sections. CA1 and CA2-3 volumes were summed to form the CA volume; CA4 and GCMLDG volumes were summed to form the CA4-dentate gyrus volume; and subiculum, presubiculum, and parasubiculum volumes were summed to form the subicular complex volume. CA: cornu ammonis; GCMLDG: the molecular and granule cell layers of the dentate gyrus; HATA: hippocampal-amygdala transition area.

Assessment of cognitive domains

Cognitive functioning was measured at baseline and annually for three years using a formal detailed neuropsychological test battery (National Institute of Neurological Disorders and Stroke–Canadian Stroke Network battery) that has been locally validated in Singapore.⁴⁴ The complete battery includes a 60-min protocol, which assesses the following six cognitive domains:

Attention: Digit span forward and backward;

Executive function: Verbal fluency, Color trail test A and B;

Language: 15-item modified Boston naming test;

Memory: Rey complex figure test – Immediate/delayed recall and recognition, Hopkins verbal learning test – Immediate/delayed recall and recognition, Picture recall and WMS-R visual reproduction;

Visuospatial function: Rey complex figure test – Copy; and

Visuomotor speed: Symbol digit modalities test.

The individual test raw scores were transformed to standardized z-scores using the mean and SD of controls (i.e., NCI) for each score and averaged to obtain the z-scores for each cognitive domain. The control group used consisted of patients in the memory clinic cohort who had presented to the clinic with subjective cognitive complaints but were determined on neuropsychological testing to have NCI, as well as spouses and relatives of the cases. The standardized global score was computed by averaging the domain z-scores and standardizing using the mean and SD of the controls. Similarly, the standardized global and domain cognitive z-scores for each follow-up visit were obtained using the means and SDs of the control group at baseline.

Statistical analysis

Characteristics of the study population were summarized using mean and SD for continuous variables, and numbers and percentages for categorical variables. Differences in the baseline characteristics of patients with different baseline diagnoses were assessed using chi-square tests for categorical variables, and one-way ANOVA for continuous variables. False Discovery Rate (FDR) correction was applied to correct for multiple comparisons. To identify specific group comparisons (e.g., NCI versus CIND versus dementia) driving the differences, post-hoc Tukey's HSD tests were conducted for continuous variables that remained significant after FDR correction, while pairwise chi-square tests were carried out for categorical variables.

The associations between baseline hippocampal subfield volumes and cognitive domains were analyzed using multiple linear regression models while the associations between baseline hippocampal subfield volumes and cognitive decline from baseline to the third follow-up were analyzed using linear regression models with generalized estimating equation (GEE) to account for confounding and the correlation between repeated measurements taken during each follow-up year. A first-order autoregressive correlation structure was specified, and the sandwich estimator of the variation was used to obtain a robust standard error for the estimates. To ascertain that associations were not driven by covariates included in the model, we first utilized a simple model adjusting for age, sex, years of education, and baseline cognitive diagnosis (Model 1). We then further adjusted for the presence of lacunar infarcts, the presence of cortical infarcts, white matter hyperintensities volume, and intracranial volume (Model 2).

To investigate whether the effect of baseline hippocampal subfield volumes on cognitive decline was different between each follow-up time, we included an interaction term between hippocampal subfield volumes and the follow-up time variable. The Wald test was used to assess whether the interaction term was a significant factor. For cognitive domains with significant interaction, we assessed the effect of hippocampal subfield volumes on the cognitive score at each follow-up visit.

The associations between baseline hippocampal subfield volumes and cognitive diagnosis were analyzed using multiple logistic regression models with NCI as the reference group. To investigate the effect of baseline hippocampal subfield volumes on incident dementia, multiple logistic regression models were computed to identify the odds of incident dementia in patients without dementia at baseline. For the analyses on the odds of dementia and incident dementia, the signs of hippocampal subfield volumes were reversed such that smaller hippocampal volumes were associated with a higher odds of dementia or incident dementia. In Model 1, we adjusted for age, sex, and years of education. In Model 2, we further adjusted for the presence of lacunar infarcts, the presence of cortical infarcts, white matter hyperintensities volume, and intracranial volume. To account for multiple testing, FDR and Bonferroni corrections were applied, and a p-value of <0.05/7/8 ≈ 0.0009 was considered statistically significant. Statistical analyses were performed using R: A Language and Environment for Statistical Computing (version 1.1.463, R Foundation for Statistical Computing). The “geepack” package was used for computing the regression models with GEE.

Results

Characteristics of the study population

Characteristics of the study population are presented in Table 1 according to their baseline cognitive diagnoses. The mean age was 73.2 (±7.8) years and 237 (44.4%) patients were male. A total of 107 (20.0%) patients were diagnosed as NCI, 222 (41.6%) as CIND, and 205 (38.4%) as dementia. After FDR correction, cognitive diagnoses were significantly different with respect to age, education, lacunar infarcts, cortical infarcts, WMH, and hippocampal volume on univariate analysis. Cognitive diagnoses were significantly associated with all hippocampal subfield volumes on univariate analysis. Post-hoc tests revealed significant differences between all subgroups on all variables except for age, fissure size, lacunar infarcts, and cortical infarcts (Supplemental Tables 1 and 2).

Table 1.

Demographic and clinical characteristics of the memory clinic cohort stratified by baseline cognitive diagnosis (n = 534).

Variable	No cognitive impairment (n = 107)	Cognitive impairment no dementia (n = 222)	Dementia (n = 205)	Total (n = 534)	p
Age (y), mean (SD)	69.6 (6.6)	73.1 (7.9)	75.3 (7.6)	73.2 (7.8)	<0.001
Gender, male	51 (47.7)	107 (48.2)	79 (38.5)	237 (44.4)	0.112
Race, Chinese	98 (91.6)	188 (84.7)	167 (81.5)	453 (84.8)	0.073
Education (y), mean (SD)	9.8 (5.2)	7.4 (4.8)	4.8 (4.4)	6.9 (5.1)	<0.001
Cardiovascular disease	9 (8.4)	16 (7.2)	23 (11.2)	48 (9.0)	0.361
Presence of lacunar infarct	13 (12.2)	63 (28.4)	73 (35.6)	149 (27.9)	<0.001
Presence of cortical infarct	6 (5.6)	24 (10.8)	45 (22.0)	75 (14.0)	<0.001
Log of white matter hyperintensity volume (ml), mean (SD)	1.1 (0.9)	1.6 (1.1)	2.1 (1.2)	1.7 (1.2)	<0.001
Cornu ammonis size (ml), mean (SD)	0.80 (0.09)	0.73 (0.12)	0.64 (0.13)	0.71 (0.13)	<0.001
CA4-dentate gyrus size (ml), mean (SD)	0.50 (0.06)	0.45 (0.07)	0.39 (0.07)	0.44 (0.08)	<0.001
Hippocampus-amygdala transition area size (ml), mean (SD)	0.05 (0.01)	0.04 (0.01)	0.04 (0.01)	0.04 (0.01)	<0.001
Molecular layer size (ml), mean (SD)	0.52 (0.06)	0.46 (0.07)	0.40 (0.08)	0.45 (0.08)	<0.001
Subicular complex size (ml), mean (SD)	0.75 (0.08)	0.68 (0.11)	0.61 (0.12)	0.67 (0.12)	<0.001
Fissure size (ml), mean (SD)	0.17 (0.03)	0.16 (0.03)	0.15 (0.03)	0.16 (0.03)	<0.001
Fimbria size (ml), mean (SD)	0.06 (0.01)	0.05 (0.02)	0.04 (0.02)	0.05 (0.02)	<0.001
Tail size (ml), mean (SD)	0.53 (0.08)	0.50 (0.08)	0.43 (0.08)	0.48 (0.09)	<0.001
Total hippocampal size (ml), mean (SD)	3.37 (0.34)	3.08 (0.45)	2.69 (0.49)	2.99 (0.52)	<0.001
Intracranial volume (ml), mean (SD)	1100.2 (119.1)	1089.8 (112.2)	1092.8 (115.2)	1093.0 (114.6)	0.741

ml: milliliters; SD: standard deviation

Association of hippocampal subfield volumes with cognitive domain scores at baseline

The associations between hippocampal subfield volumes with cognitive domain scores at baseline after adjusting for age, sex, race, years of education, baseline cognitive diagnosis, cardiovascular disease, lacunar infarct, cortical infarct, white matter hyperintensities volume, and intracranial volume, were presented in Figure 3A. The CA4-dentate gyrus (β: 0.24, 95% CI: 0.12–0.36), molecular layer (β: 0.25, 95% CI: 0.12–0.37), and fimbria (β: 0.29, 95% CI: 0.16–0.41) were significantly associated with global cognition (all p < 0.0009). For cognitive domains, there was a significant association between the fissure and language. The cornu ammonis, CA4-dentate gyrus, molecular layer, subicular complex, fimbria, and tail were associated with memory; and the CA4-dentate gyrus, molecular layer, subicular complex, and fimbria with visuospatial skills (all p < 0.0009).

Figure 3.

Heatmap of associations between baseline hippocampal subfield volumes and cognitive scores. Note: Heatmap of associations between hippocampal subfield volumes and (A) baseline cognitive scores and (B) cognitive scores over time on global cognition and six cognitive domains. Models are adjusted for age, sex, education, baseline cognitive diagnosis, lacunar infarct, cortical infarct, white matter hyperintensities volume, and intracranial volume. Colors correspond to beta estimates (red denotes positive associations and blue denotes negative associations). Statistical significance after FDR and Bonferroni corrections is indicated by an asterisk: *p < 0.0009.

Association of hippocampal subfield volumes with cognitive domain scores over three years of follow up

The associations between hippocampal subfield volumes with cognitive domain scores over time after adjusting for confounders and the FDR and Bonferroni corrections were presented in Figure 3B and Supplemental Table 3. The cornu ammonis (β: 0.30, 95% CI: 0.17–0.44), CA4-dentate gyrus (β: 0.32, 95% CI: 0.19–0.46), HATA (β: 0.29, 95% CI: 0.14–0.43), molecular layer (β: 0.34, 95% CI: 0.18–0.49), subicular complex (β: 0.30, 95% CI: 0.14–0.46), and fimbria (β: 0.26, 95% CI: 0.13–0.40) were significantly associated with global cognition (all p < 0.0009). The cornu ammonis, CA4-dentate gyrus, HATA, molecular layer, subicular complex, fimbria, and tail were significantly associated with memory (all p < 0.0009).

The interaction effects between hippocampal subfield volumes and follow-up time with cognitive domain scores after adjusting for confounders and the FDR and Bonferroni correction were presented in Figure 4 and Supplemental Table 4. For all hippocampal subfield volumes apart from the hippocampal fissure, there were significant interactions with follow-up time on global cognition, language, and memory such that with increasing time, atrophy in hippocampal subfields were associated with a larger decline in cognitive domain scores. There was a significant interaction between fimbria volumes and follow-up time on visuomotor speed. There were no significant interaction effects between hippocampal subfield volumes and follow-up time on attention, executive function and visuospatial skills.

Figure 4.

Interaction between baseline hippocampal subfield volumes and time on cognitive scores. Note: Estimates of significant interaction effects between hippocampal subfield volumes and time on cognitive scores using generalized estimating equations, after adjusting for age, gender, education, baseline cognitive diagnosis, lacunar infarct, cortical infarct, white matter hyperintensities volume, and intracranial volume. Colors indicate hippocampal subfields.

Association of hippocampal subfield volumes with cognitive diagnosis at baseline and incident dementia

30 patients were diagnosed with incident dementia at follow up. The vast majority were diagnosed with AD (n = 28), while the remaining two patients were diagnosed with vascular dementia. The association between hippocampal subfield volumes with baseline cognitive diagnoses and incident dementia were reported in Table 2. All hippocampal subfield volumes were associated with diagnosis of dementia at baseline, while the CA4-dentate gyrus, molecular layer, subicular complex, and fimbria volumes were associated with diagnosis of CIND at baseline. Furthermore, molecular layer volume was significantly associated with incident dementia after two-year follow-up. These associations were significant even after adjusting for age, gender, race, education, cardiovascular disease, lacunar infarct, cortical infarct, white matter hyperintensities, and intracranial volume.

Table 2.

Logistic regression models for the association between baseline hippocampal subfield volume and cognitive diagnosis.

Hippocampal Subfields	Cognitive Diagnosis^{† ¶}	Model 1^{‡ ¶}		Model 2^{§ ¶}
Hippocampal Subfields	Cognitive Diagnosis^{† ¶}	OR (95% CI)	p	OR (95% CI)	p
Cornu ammonis	CIND	1.57 (1.23; 2.02)	0.001	1.58 (1.23; 2.06)	0.001
	Dementia	2.92 (2.19; 4.02)	<0.001	3.48 (2.51; 5.06)	<0.001
	Incident Dementia ^#	1.96 (1.32; 2.98)	0.002	2.11 (1.41; 3.27)	0.001
CA4-dentate gyrus	CIND	1.64 (1.30; 2.11)	<0.001	1.69 (1.31; 2.20)	<0.001
	Dementia	3.65 (2.66; 5.21)	<0.001	4.28 (2.98; 6.54)	<0.001
	Incident Dementia ^#	1.85 (1.27; 2.75)	0.003	2.02 (1.37; 3.07)	0.001
Hippocampus-Amygdala Transition Area	CIND	1.52 (1.21; 1.94)	0.001	1.50 (1.18; 1.93)	0.002
	Dementia	2.38 (1.81; 3.21)	<0.001	2.61 (1.95; 3.62)	<0.001
	Incident Dementia ^#	1.30 (0.88; 1.98)	0.209	1.38 (0.92; 2.12)	0.130
Molecular layer	CIND	1.82 (1.41; 2.38)	<0.001	1.84 (1.41; 2.44)	<0.001
	Dementia	3.58 (2.62; 5.15)	<0.001	4.06 (2.85; 6.13)	<0.001
	Incident Dementia ^#	2.09 (1.41; 3.16)	0.001	2.26 (1.50; 3.50)	<0.001
Subicular complex	CIND	1.63 (1.29; 2.08)	<0.001	1.70 (1.33; 2.21)	<0.001
	Dementia	2.96 (2.21; 4.11)	<0.001	3.53 (2.53; 5.18)	<0.001
	Incident Dementia ^#	1.68 (1.15; 2.49)	0.010	1.86 (1.25; 2.84)	0.004
Fissure	CIND	1.12 (0.88; 1.43)	0.352	1.20 (0.93; 1.55)	0.165
	Dementia	1.74 (1.32; 2.32)	<0.001	2.11 (1.54; 2.98)	<0.001
	Incident Dementia ^#	1.34 (0.87; 2.12)	0.209	1.52 (0.96; 2.52)	0.094
Fimbria	CIND	1.94 (1.52; 2.51)	<0.001	1.89 (1.47; 2.47)	<0.001
	Dementia	4.21 (2.96; 6.32)	<0.001	5.30 (3.46; 8.78)	<0.001
	Incident Dementia ^#	1.85 (1.25; 2.81)	0.003	1.78 (1.19; 2.72)	0.007
Tail	CIND	1.35 (1.03; 1.78)	0.038	1.36 (1.02; 1.82)	0.040
	Dementia	3.14 (2.28; 4.46)	<0.001	3.42 (2.41; 5.05)	<0.001
	Incident Dementia ^#	2.02 (1.29; 3.23)	0.003	2.33 (1.44; 3.90)	0.001

CA4: cornu ammonis 4; CI: confidence interval; CIND: cognitive impairment with no dementia; OR: odds ratio

^†

NCI is the reference level

^‡

Model 1: Adjusted for age, gender, education

^§

Model 2: Adjusted for age, gender, education, lacunar infarct, cortical infarct, white matter hyperintensities volume, intracranial volume

^¶

Negative and positive signs of standardized hippocampal subfield volumes reversed such that odds of CIND and dementia diagnoses increases as hippocampal subfield volume decreases

Incident dementia after two years

Discussion

In a memory clinic cohort, we found that decreased molecular layer size was significantly associated with increased risk of dementia after two years (OR 2.26) after adjusting for confounders. Furthermore, on annual neuropsychological testing in specific cognitive domains over three years of follow up, all hippocampal subfields other than the hippocampal fissure were associated with memory, and all hippocampal subfields apart from the hippocampal fissure and tail were significantly associated with global cognition, even after adjusting for confounders. All these associations survived correction for multiple comparisons.

Our findings corroborate existing evidence that hippocampal subfield atrophy were associated with impairment in specific cognitive domains and can predict incident dementia.^8–12,25 Specifically, we found that hippocampal atrophy in all subfields except for the hippocampal fissure were significantly associated with poorer memory over time, and atrophy in all subfields apart from the hippocampal fissure and tail were associated with poorer global cognition after three years of follow up.^27,45 Of the hippocampal subfields, studies have suggested stronger associations of atrophy of the subicular complex, cornu ammonis, and molecular layer with the development of dementia.^8,46 In our study, only the association between molecular layer volume and incident dementia survived FDR correction and correction for multiple testing. In addition to hippocampal subfields, the microstructure of the entorhinal cortex has also been identified as an early biomarker for dementia.⁴⁷

The CA4-dentate gyrus, molecular layer, subicular complex, and fimbria were also associated with baseline visuospatial skills, but these associations did not persist over time. These hippocampal subfields were also associated with a CIND diagnosis at baseline, and the associations with baseline visuospatial skills may reflect the role of the fimbria in spatial learning and the involvement of the CA4-dentate gyrus, molecular layer, and subicular complex in the perforant pathway, which is essential for spatial information processing.^19,47,48 Our study suggested that smaller hippocampal subfields at baseline did not predict a decline in visuospatial skills. However, we note that the associations between hippocampal subfield volumes and cognitive domain scores over time were largely similar to that with baseline cognitive domain scores, and that differences in statistical significance could be explained by differences in the sample size. Future studies would be valuable to further clarify our understanding of the associations between hippocampal subfield volumes and cognitive domain scores at baseline and over time.

Our findings support the hypothesis that different subfields of the hippocampus may contribute to different trajectories of cognitive decline in specific cognitive domains.^24,25 Specifically, hippocampal atrophy of the cornu ammonis, CA4-dentate gyrus, and molecular layer may first manifest with cognitive impairment in memory before other subfields of the hippocampus, and molecular layer volume may be an early biomarker of incident dementia over time. This hypothesis may be tested using serial imaging of the hippocampal subfields in a cohort of patients with early cognitive impairment who subsequently develop dementia over time.

Further analysis showed that for all hippocampal subfield volumes apart from the hippocampal fissure, there were significant interactions with follow-up time on global cognition such that the positive association between hippocampal subfield size and cognitive scores increased at the second and third year of follow up. This was also found for the language and memory cognitive domains, even after adjusting for confounding and the FDR and Bonferroni corrections. This suggests that over time, atrophy in hippocampal subfields was associated with a larger decline in cognitive domain scores independent of other clinical and imaging biomarkers of cognition. The lack of significant interactions between the hippocampal subfield volumes and follow-up time on attention suggests that their significant interactions with memory likely reflect true increases in association with memory and learning over time,⁴⁹ rather than being influenced by attention.⁵⁰ This was important to examine considering the role of attention in encoding new memories.⁵¹ Our findings are consistent with existing evidence that primary memory impairment is mainly driven by hippocampal atrophy, while attentional impairments are more strongly associated with frontal lobe atrophy.^52,53

To our knowledge, this is the first study to evaluate the independent effects of hippocampal subfield volumes with changes in specific cognitive domains over three years of follow up, as well as the risk of dementia in a prospective memory clinic cohort. The strengths of our study include the use of an extensive neuropsychological test battery to quantify specific cognitive domains over three years of follow up, as well as a detailed neuroimaging protocol for quantification of hippocampal subfields and other MRI-based markers of cerebrovascular disease. We were able to account for confounders of cognitive impairment and dementia in our analyses, as well as showed statistically significant effects even after the FDR and Bonferroni corrections. Utilizing a hierarchical modeling approach enabled us to ascertain that the observed associations were not driven by the covariates included in the models. Considering that higher-order cognitive functions such as memory are supported by fundamental cognitive functions such as attention and processing speed,^50,54 investigating the associations between hippocampal subfield volumes and changes in various key cognitive domains allowed us to clarify the effect of smaller subfield volumes on decline in specific areas of cognition.

The limitations of our study were that we were unable to conduct subgroup analysis based on cognitive diagnosis to examine the longitudinal associations of hippocampal subfield atrophy in the cohort of patients with NCI as the study was insufficiently powered to do so. As our patients were recruited from a memory clinic, many patients were cognitively impaired at baseline, which might have led to selection bias and limited the generalizability of our findings to the general population. However, we were able to account for baseline cognitive diagnosis as well as risk factors for dementia in our analysis, and the longitudinal associations identified in this cohort were able to provide new insights into the role of hippocampal subfield atrophy on specific cognitive domains and incident dementia over time.

Conclusion

As decreased size in the molecular layer were significantly associated with increased risk of incident dementia after two years, our findings suggest that hippocampal atrophy in this subfield may be an early biomarker of incident dementia over time. Hippocampal atrophy of the cornu ammonis, CA4-dentate gyrus, and molecular layer may manifest with cognitive impairment in memory before other subfields of the hippocampus. Furthermore, all hippocampal subfields other than the hippocampal fissure were associated with memory on annual neuropsychological testing over three years of follow-up. Further research demonstrating sequential atrophy in hippocampal subfields, and the biological role of hippocampal subfields in specific cognitive domains are required to confirm these findings.

Supplemental Material

sj-docx-1-alz-10.1177_13872877251329574 - Supplemental material for The effect of hippocampal subfield volumes on cognitive decline and incident dementia in a memory clinic cohort

Supplemental material, sj-docx-1-alz-10.1177_13872877251329574 for The effect of hippocampal subfield volumes on cognitive decline and incident dementia in a memory clinic cohort by Mervyn JR Lim, Jaclyn Tan, Caroline Robert, Wei Ying Tan, Narayanaswamy Venketasubramanian, Christopher Chen and Saima Hilal in Journal of Alzheimer's Disease

Footnotes

Acknowledgments

The authors have no acknowledgments to report.

ORCID iDs

Mervyn JR Lim

Jaclyn Tan

Caroline Robert

Narayanaswamy Venketasubramanian

Christopher Chen

Saima Hilal

Ethical considerations

Institutional ethics approval (National Healthcare Group Domain Specific Review Board) was obtained prior to the conduct of the study.

Consent to participate

Written informed consent was obtained for all human subjects.

Consent for publication

Not applicable

Author contributions

Mervyn JR Lim (Conceptualization; Formal analysis; Writing – original draft; Writing – review & editing); Jaclyn Tan (Formal analysis; Writing – original draft; Writing – review & editing); Caroline Robert (Investigation; Writing – review & editing); Wei Ying Tan (Data curation; Investigation; Methodology; Writing – review & editing); Narayanaswamy Venketasubramanian (Investigation; Writing – review & editing); Christopher Chen (Supervision; Writing – review & editing); Saima Hilal (Conceptualization; Supervision; Writing – review & editing).

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Medical Research Council (NMRC) of Singapore [grant numbers NMRC/CG/NUHS/2010, NMRC/CG/013/2013, NMRC/CIRG/1485/2018], Center Grant SEED Funding [grant number R-184-000-294-511], the Yong Loo Lin School of Medicine Aspiration Fund, the National Medical Research Council Singapore, Transition Award [grant number R-608-000-342-213], and the Ministry of Education, Academic Research Fund Tier 1 [grant number A-0006106-00-00] and Academic Health Programme (grant number A-8002831-00-00).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability

The data that support the findings of this study are available on request from the corresponding author.

Supplemental material

Supplemental material for this article is available online.

References

Eichenbaum

Cohen

. Can we reconcile the declarative memory and spatial navigation views on hippocampal function? Neuron 2014; 83: 764–770.

O'Shea

Cohen

Porges

, et al. Cognitive aging and the hippocampus in older adults. Front Aging Neurosci 2016; 8: 298.

Pini

Pievani

Bocchetta

, et al. Brain atrophy in Alzheimer's disease and aging. Ageing Res Rev 2016; 30: 25–48.

Coupe

Manjon

Mansencal

, et al. Hippocampal-amygdalo-ventricular atrophy score: Alzheimer disease detection using normative and pathological lifespan models. Hum Brain Mapp 2022; 43: 3270–3282.

Jack

Jr. Bennett

Blennow

, et al. NIA-AA Research Framework: toward a biological definition of Alzheimer's disease. Alzheimers Dement 2018; 14: 535–562.

Fjell

McEvoy

Holland

, et al. What is normal in normal aging? Effects of aging, amyloid and Alzheimer's disease on the cerebral cortex and the hippocampus. Prog Neurobiol 2014; 117: 20–40.

van der Flier

Scheltens

. Alzheimer disease: hippocampal volume loss and Alzheimer disease progression. Nat Rev Neurol 2009; 5: 361–362.

Evans

Adams

HHH

Licher

, et al. Subregional volumes of the hippocampus in relation to cognitive function and risk of dementia. Neuroimage 2018; 178: 129–135.

Mueller

Chao

Berman

, et al. Evidence for functional specialization of hippocampal subfields detected by MR subfield volumetry on high resolution images at 4T. Neuroimage 2011; 56: 851–857.

10.

Suthana

Donix

Wozny

, et al. High-resolution 7T fMRI of human hippocampal subfields during associative learning. J Cogn Neurosci 2015; 27: 1194–1206.

11.

Shavitt

Johnson

INS

Batistuzzo

. Hippocampal formation volume, its subregions, and its specific contributions to visuospatial memory tasks. Braz J Med Biol Res 2020; 53: e9481.

12.

Kesner

. An analysis of the dentate gyrus function. Behav Brain Res 2013; 254: 1–7.

13.

Danieli

Guyon

Bethus

. Episodic memory formation: a review of complex hippocampus input pathways. Prog Neuropsychopharmacol Biol Psychiatry 2023; 126: 110757.

14.

Granger

Colon-Perez

Larson

, et al. Reduced structural connectivity of the medial temporal lobe including the perforant path is associated with aging and verbal memory impairment. Neurobiol Aging 2023; 121: 119–128.

15.

Jacobs

HIL

Augustinack

Schultz

, et al. The presubiculum links incipient amyloid and tau pathology to memory function in older persons. Neurology 2020; 94: e1916–e1928.

16.

Zeineh

Palomero-Gallagher

Axer

, et al. Direct visualization and mapping of the spatial course of fiber tracts at microscopic resolution in the human hippocampus. Cereb Cortex 2017; 27: 1779–1794.

17.

Moorthi

Premkumar

Priyanka

, et al. Pathological changes in hippocampal neuronal circuits underlie age-associated neurodegeneration and memory loss: positive clue toward SAD. Neuroscience 2015; 301: 90–105.

18.

Mankin

Aghajan

Schuette

, et al. Stimulation of the right entorhinal white matter enhances visual memory encoding in humans. Brain Stimul 2021; 14: 131–140.

19.

Hunsaker

Mooy

Swift

, et al. Dissociations of the medial and lateral perforant path projections into dorsal DG, CA3, and CA1 for spatial and nonspatial (visual object) information processing. Behav Neurosci 2007; 121: 742–750.

20.

Haam

Gunin

Wilson

, et al. Entorhinal cortical delta oscillations drive memory consolidation. Cell Rep 2023; 42: 113267.

21.

Llorens-Martin

Blazquez-Llorca

Benavides-Piccione

, et al. Selective alterations of neurons and circuits related to early memory loss in Alzheimer's disease. Front Neuroanat 2014; 8: 38.

22.

Shohamy

Turk-Browne

. Mechanisms for widespread hippocampal involvement in cognition. J Exp Psychol Gen 2013; 142: 1159–1170.

23.

Olsen

Moses

Riggs

, et al. The hippocampus supports multiple cognitive processes through relational binding and comparison. Front Hum Neurosci 2012; 6: 46.

24.

Smits

van Harten

Pijnenburg

, et al. Trajectories of cognitive decline in different types of dementia. Psychol Med 2015; 45: 1051–1059.

25.

Apostolova

Mosconi

Thompson

, et al. Subregional hippocampal atrophy predicts Alzheimer's dementia in the cognitively normal. Neurobiol Aging 2010; 31: 1077–1088.

26.

Tsalouchidou

Muller

Belke

, et al. Verbal memory depends on structural hippocampal subfield volume. Front Neurol 2023; 14: 1209941.

27.

Huang

Wang

, et al. Differential associations of visual memory with hippocampal subfields in subjective cognitive decline and amnestic mild cognitive impairment. BMC Geriatr 2022; 22: 153.

28.

Wannan

CMJ

Cropley

Chakravarty

, et al. Hippocampal subfields and visuospatial associative memory across stages of schizophrenia-spectrum disorder. Psychol Med 2019; 49: 2452–2462.

29.

Izzo

Andreassen

Westlye

, et al. The association between hippocampal subfield volumes in mild cognitive impairment and conversion to Alzheimer's disease. Brain Res 2020; 1728: 146591.

30.

Hilal

Chai

van Veluw

, et al. Association between subclinical cardiac biomarkers and clinically manifest cardiac diseases with cortical cerebral microinfarcts. JAMA Neurol 2017; 74: 403–410.

31.

Lim

MJR

Tan

Gyanwali

, et al. The associations between intracranial stenosis, brain amyloid-beta, and cognition in a memory clinic sample. Alzheimer Dis Assoc Disord 2022; 36: 327–334.

32.

Hilal

Saini

Tan

, et al. Intracranial stenosis, cerebrovascular diseases, and cognitive impairment in Chinese. Alzheimer Dis Assoc Disord 2015; 29: 12–17.

33.

Saridin

Hilal

Villaraza

, et al. Brain amyloid beta, cerebral small vessel disease, and cognition: a memory clinic study. Neurology 2020; 95: e2845–e2853.

34.

Lim

MJR

Tan

Gyanwali

, et al. The effect of intracranial stenosis on cognitive decline in a memory clinic cohort. Eur J Neurol 2021; 28: 1829–1839.

35.

de Boer

Vrooman

Ikram

, et al. Accuracy and reproducibility study of automatic MRI brain tissue segmentation methods. Neuroimage 2010; 51: 1047–1056.

36.

Vrooman

Cocosco

van der Lijn

, et al. Multi-spectral brain tissue segmentation using automatically trained k-Nearest-Neighbor classification. Neuroimage 2007; 37: 71–81.

37.

Ikram

Vrooman

Vernooij

, et al. Brain tissue volumes in the general elderly population. The Rotterdam Scan Study. Neurobiol Aging 2008; 29: 882–890.

38.

Hilal

Saini

Tan

, et al. Cerebral microbleeds and cognition: the epidemiology of dementia in Singapore study. Alzheimer Dis Assoc Disord 2014; 28: 106–112.

39.

Fischl

Salat

Busa

, et al. Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron 2002; 33: 341–355.

40.

Maurus

Roell

Keeser

, et al. Fitness is positively associated with hippocampal formation subfield volumes in schizophrenia: a multiparametric magnetic resonance imaging study. Transl Psychiatry 2022; 12: 388.

41.

Princich

Donnelly-Kehoe

Deleglise

, et al. Diagnostic performance of MRI volumetry in epilepsy patients with hippocampal sclerosis supported through a random forest automatic classification algorithm. Front Neurol 2021; 12: 613967.

42.

Bodi

Polgar

Kiss

, et al. Reduced volumes of the CA1 and CA4-dentate gyrus hippocampal subfields in systemic lupus erythematosus. Lupus 2017; 26: 1378–1382.

43.

Beale

Broyd

Chye

, et al. Prolonged cannabidiol treatment effects on hippocampal subfield volumes in current cannabis users. Cannabis Cannabinoid Res 2018; 3: 94–107.

44.

Chan

Hilal

, et al. The diagnostic utility of the NINDS-CSN neuropsychological battery in memory clinics. Dement Geriatr Cogn Dis Extra 2016; 6: 276–282.

45.

Wang

, et al. Volume changes of hippocampal and amygdala subfields in patients with mild cognitive impairment and Alzheimer's disease. Acta Neurol Belg 2023; 123: 1381–1393.

46.

Zhao

Wang

Yin

, et al. Trajectories of the hippocampal subfields atrophy in the Alzheimer's disease: a structural imaging study. Front Neuroinform 2019; 13: 13.

47.

Uchida

Onda

Hou

, et al. Microstructural neurodegeneration of the entorhinal-hippocampus pathway along the Alzheimer's disease continuum. J Alzheimers Dis 2023; 95: 1107–1117.

48.

Dahmani

Courcot

Near

, et al. Fimbria-fornix volume is associated with spatial memory and olfactory identification in humans. Front Syst Neurosci 2019; 13: 87.

49.

Tamnes

Walhovd

Engvig

, et al. Regional hippocampal volumes and development predict learning and memory. Dev Neurosci 2014; 36: 161–174.

50.

Saunders

Summers

. Attention and working memory deficits in mild cognitive impairment. J Clin Exp Neuropsychol 2010; 32: 350–357.

51.

Cowan

Bao

Bishop-Chrzanowski

, et al. The relation between attention and memory. Annu Rev Psychol 2024; 75: 183–214.

52.

Nordahl

Ranganath

Yonelinas

, et al. Different mechanisms of episodic memory failure in mild cognitive impairment. Neuropsychologia 2005; 43: 1688–1697.

53.

Bruck

Kurki

Kaasinen

, et al. Hippocampal and prefrontal atrophy in patients with early non-demented Parkinson's disease is related to cognitive impairment. J Neurol Neurosurg Psychiatry 2004; 75: 1467–1469.

54.

MacDonald

Almor

Henderson

, et al. Assessing working memory and language comprehension in Alzheimer’s disease. Brain Lang 2001; 78: 17–42.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.05 MB