Advertisement

WHAT HAPPENS TO THE HIPPOCAMPUS 12 MONTHS AFTER TRAINING? LONGITUDINAL LINEAR MIXED-EFFECTS MODEL ANALYSIS OF MILD COGNITIVE IMPAIRMENT IN THE SMART TRIAL

      Background

      Cognitive and physical exercise may reduce risk of dementia in mild cognitive impairment (MCI) but the underlying mechanisms are poorly understood. The SMART is a longitudinal randomised controlled trial that compares the benefits of isolated and combined progressive resistance and computerised cognitive training in MCI. Previously, we revealed therapeutically relevant structural and functional brain changes immediately after training cessation1, however, no effects were found on hippocampal structure. Here, for the first time, we investigate the ongoing impact of training on hippocampal anatomy 12-months after cessation using linear mixed effects (LME) models to account for imperfect timing and missing data.

      Methods

      Eighty six community-dwelling participants aged ≥55 with MCI were randomised into 4 training groups; 1- Combined computerised cognitive and progressive resistance training (CT+PRT), 2- PRT and Sham CT (PRT), 3- Sham PRT and CT (CT), 4- Double Sham (DS). Training consisted of 2x1.5hours/week for 6-months. Cognitive and MRI assessments were carried out at baseline (BL), 6-months (F1) (directly after training) and 18-months (F2) from BL. Whole-brain T13DTFE MR images were automatically processed with the longitudinal FreeSurfer analysis pipeline. Longitudinal hippocampal volume was analysed with a freely available LME Matlab tool, modelled as % of BL volume. A linear time x group interaction was selected as the main contrast of interest in a comprehensive model controlling for covariates (sex, education, age).

      Results

      Five alternate hypotheses tested for group x time interactions between BL->F1 and BL->F1->F2 (Table 1). LME models showed significant differences in left (but not right) hippocampal atrophy rates for either training intervention compared to DS across the entire 18-month follow-up period. There were no differences in hippocampal trajectories between PRT or CT, nor between combined training and the DS.

      Conclusions

      Meta-analysis in MCI and Alzheimer’s confirm a faster rate of atrophy in the left hippocampus compared to the right2. Our findings of preserved left but not right hippocampal volume due to training may therefore have implications for combating neurodegeneration. Further work is needed to determine if such training-related benefits are linked to improved memory and cognitive outcomes long term. References: 1-C. Suo, et al. Molecular Psychiatry. 2016.21:1633–1642 2-F. Shi et al. Hippocampus. 2009. 19:1055–1064.
      Figure thumbnail fx1
      Figure 1Render of hippocampi from FreeSurfer segmentation (left). Lowess plot of left hippocampal volume as a percentage of ICV (right). The lowess plot shows predicted values of left hippocampal volume from LME model. Sliding window data fraction was set to 0.9.
      Table 1
      Significance difference in rate of change betweenBL-FU1 pDFF valueBL-FU1-FU2 pDFF value
      H1 - DS and all training groups0.34975.30.890.025155.14.70
      H2 - DS and CT0.11940.11.840.02784.45.10
      H3 - DS and PRT0.18336.12.540.02780.85.04
      H4 - PRT and CT0.36939.00.830.58572.20.30
      H5 - PRT+CT and DS0.57339.10.320.11482.92.02