Hippocampal CA3-dentate gyrus volume uniquely linked to improvement in associative memory from childhood to adulthood

Abstract

Associative memory develops into adulthood and critically depends on the hippocampus. The hippocampus is a complex structure composed of subfields that are functionally-distinct, and anterior-posterior divisions along the length of the hippocampal horizontal axis that may also differ by cognitive correlates. Although each of these aspects has been considered independently, here we evaluate their relative contributions as correlates of age-related improvement in memory. Volumes of hippocampal subfields (subiculum, CA1-2, CA3-dentate gyrus) and anterior-posterior divisions (hippocampal head, body, tail) were manually segmented from high-resolution images in a sample of healthy participants (age 8-25 years). Adults had smaller CA3-dentate gyrus volume as compared to children, which accounted for 67% of the indirect effect of age predicting better associative memory via hippocampal volumes. Whereas hippocampal body volume demonstrated non-linear age differences, larger hippocampal body volume was weakly related to better associative memory only when accounting for the mutual correlation with subfields measured within that region. Thus, typical development of associative memory was largely explained by age-related differences in CA3-dentate gyrus.

Keywords: Associative memory; Children; Dentate gyrus; Hippocampal body; Hippocampal subfields.

Figure 1

Example manual tracing of hippocampal subfields and head, body tail on T2 proton-density weighted turbo spin echo images. Image intensities are inverted. (A) Example of manual tracing of hippocampal head (green), body (orange), and tail (purple). (B) An example sagittal T1-weighted image that displays the approximate range of head (green), body (orange) and tail (purple) measurements made from the T2-weighted images. The white bracket, labeled “C” indicates approximately the 3 slices on which the subfields were segmented in the anterior Hc body, beginning posterior to the uncal apex. (C) Example of manual tracing of the hippocampal subfields were made on three contiguous slices of the anterior body on T2 proton-density weighted turbo spin echo images: subiculum (yellow), CA1-2 (blue), CA3-dentate gyrus (red). Additional regions are labeled for anatomical reference: Am—amygdala; EC—entorhinal cortex; Fx—column of fornix; MB—mammillary body; PhG—parahippocampal gyrus; Pu—pulvinar nucleus.

Figure 2

Age differences in volumes of hippocampal subfields and hippocampal anterior-posterior divisions. (A) CA3-dentate gyrus (age p = 0.02; age² p = 0.62; R² = 0.12), (B) CA1-2 (age p = 0.56; age² p = 0.01; R² = 0.12), (C) subiculum (age p = 0.37; ge² p = 0.39; R² = 0.04);; (D) hippocampal head (age p = 0.65; age² p = 0.11; R² = 0.05), (E) hippocampal body (age p = 0.10; age² p = 0.01; R² = 0.14), (F) hippocampal tail (age p = 0.26; age² p = 0.06; R² = 0.10). Standardized effect coefficients are reported from the latent modeling that estimated linear and quadratic age differences in all regions simultaneously, accounting for correlations among subregions. Points labeled with gray represent univariate outliers that were winsorized. All volumes were corrected for intracranial volume. For plots of age differences in hippocampal subregion volumes that were not corrected for intracranial volume, see supplementary material Figure S1.

Figure 3

Age differences in memory recognition (d′) for item (age p = 0.68; age² p = 0.87; R² = 0.003) and associative pairs (age p < 0.001; age² p = 0.93; R² = 0.41). Values for memory recognition are latent estimates per each individual that were extracted from the structural equation model. Regression lines were fit. Solid line with black dots represents memory for items, and broken line with gray dots represents memory for associative pairs.

Figure 4

Primary hypothesis models testing age-related differences in hippocampal regional volumes as predicting differences in recognition memory. (A) Hippocampal subfield (subiculum, CA3-dentate gyrus, and CA1-2) volumes predicting differences in recognition memory (indirect age effect p = 0.04; R² = 0.21). (B) Hippocampal head, body and tail volumes did not predict differences in recognition memory (indirect age effect p > 0.05, non-significant paths were constrained). All coefficients are standardized. * indicates a significant effect, p < 0.05. Paths marked with broken lines indicate non-significant covariate effects.

Figure 5

Plot of the relation between age, CA3-dentate gyrus volume, and associative recognition (d′). A regression plane was fit to the three dimensional scatter plot, the plane is color coded to demonstrate the three-way interaction between age, CA3-dentate gyrus volume and associative recognition—blue corresponds to children who had large CA3-dentate gyrus volumes and poor associative recognition; yellow represents adults with smaller CA3-dentate gyrus volumes and better associative recognition. Scales for CA3-dentate gyrus volume and associative d′ are latent values for each individual extracted from the structural equation model of these effects. The indirect effect of age on associative recognition via CA3-dentate gyrus explained 67% of the total indirect effect of age on associative memory in the competing hypotheses model that included the Hc head, body, and tail volumes.