The Effect of Physical Exercise on Hippocampal
Volume and Neurogenesis
University of Amsterdam Master of Brain and Cognition
Psychology
Elena Amalie Köstler 10373071
28-06-2017 Master thesis Final Version
Dr. Filip van Opstal Dr. Timo Stein Dr. Anouk Schrantee
1. Abstract
Physical exercise seems not only beneficial for physical health but improves cognition and changes the structure of the brain. Especially the volume of the hippocampus and the neurogenesis mediating brain derived neurotrophic factor (BDNF), have shown to increase in response to aerobic exercise. However, it is not yet understood how physical exercise influences brain structure exactly and whether higher levels of exercise intensity lead to greater structural changes. In this study we scanned 47 sedentary young adults before and after a twelve-week high and low intensity exercise intervention. Surprisingly, we found no evidence of exercise induced hippocampal volume increase as has been previously reported. Rather, a small (non-significant) shrinkage in volume after exercise compared to before. Furthermore, we found a significant relationship between BDNF increase and right DG volume decrease. Our results suggest that exercise induced struc-tural changes are not as straightforward as previously thought.
The Effect of Physical Exercise on Hippocampal Volume and Neurogenesis
1. Abstract 2
2. Introduction 4
3. Materials and Methods 6
3.1 Participants and Experimental Design 6
3.2 Intervention and Procedure 7
3.2.1 Assessment of Fitness Levels (VO₂max, HR) 7
3.2.2 Exercise Intervention 7
3.2.3 Measurement of Growth Factor Levels 8
3.2.4 Magnetic Resonance Imaging: 7 Tesla high-resolution MRI 9
3.2.5 Automated Segmentation 9
3.2.5.1 Freesurfer (version 6.0) 9
3.2.5.2 Automatic Segmentation of Hippocampal Subfields (ASHS) 10 3.2.5.3 Statistical Parametric Mapping: SPM 12 10
3.3 Statistical Analysis 11
4. Results 13
4.1. Demographics and Manipulation Check 14
4.2 Cardiorespiratory Fitness and Neurogenesis 15 4.3 Cardiorespiratory Fitness and Hippocampal Volume 15
4.4 Neurogenesis and Hippocampal Volume 16
4.5 Fitness Intensities, Neurogenesis and Hippocampal Volume 18
5. Discussion 21
6. Literature 24
2. Introduction
An active physical lifestyle is beneficial not only for the body, but also for the brain. Car-diovascular fitness has, for example been associated with intelligence (Åberg et al., 2009). Physical exercise also seems to be beneficial for processing speed, episodic memory, up-dating information, and executive function (Jonasson et al., 2016). Additionally aerobic exercise improved memory performance and reversed age-related hippocampal volume loss in late adulthood (Erickson et al., 2011). Moreover, exercising regularly is advanta-geous for patients with neurodegenerative disorders as it has been shown to improve cognitive functioning of patients with dementia (Groot et al., 2016). Furthermore regu-lar physical activity reduced the risk of developing Alzheimer's disease (Buchman et al., 2012) and cognitive decline (Anderson-Hanley et al., 2012). Thus, it seems that physical exercise benefits cognitive functioning and mental health to a broad extent. However, it is not entirely clear exactly how the brain structure is influenced by regular physical exer-cise. It is also not clear yet whether structural changes are influenced by the degree of exercise intensity and if so, to what extent.
The hippocampus has been mainly the focus of research due to its exercise rela-ted potential for structural and functional changes. That is, exercising regularly increased cardiorespiratory fitness and elicited neurogenesis in rodents (Biedermann et al., 2016; Nokia et al., 2016). Neurogenesis in the adult human brain is a complex process which originates from precursor cells (stem cells and lineagedetermined progenitor cells) (Manganas et al., 2007). The precursor cells are located in the subgranular zone (SGZ) of the hippocampus and the subventricular zone (SVZ) of the lateral ventricles (Kem-permann, Wiskott, & Gage, 2004). In the hippocampus exercise induced neurogenesis mainly occurs in the anterior portion, a subfield called dentate gyrus (DG) (Sahay et al., 2011). The DG is unique to generate new neurons in mammals and in humans (Kem-permann et al., 2012; Spalding et al., 2013). The growth of these newborn cells is stimu-lated by growth factors, such as the brain-derived neurotrophic factor (BDNF). BDNF supports neuronal growth, survival and synaptic plasticity and is highly concentrated in the hippocampus (Gottmann, et al., 2009; Marlatt Potter & Lucassen, 2012). Therefore BDNF levels can serve as an indicator for neurogenesis (Wei et al., 2015). A common
method used to obtain a proxy measurement of hippocampal BDNF and neurogenesis in the human brain in vivo are therefore peripheral BDNF levels (Karege, Schwald & Cis-se, 2002).
Previous studies showed that the link between physical exercise and structural changes in the hippocampus is not straightforward. Cardiorespiratory fitness was associ-ated with hippocampal volume increase in non-demented older adults (Erickson et al., 2009). Furthermore, aerobic training in late adulthood led, for example, to an increase in hippocampal volume by 2 per cent, but to a volume decrease after only stretching exer-cises, which was also associated with the neurogenesis mediator BDNF (Erickson et al, 2011). Surprisingly, instead of a hippocampal volume increase, Wagner et al. (2015) found a decrease of 2 per cent in the right hemispheric hippocampal volume after aero-bic exercise in young male adults. Thus it seems unclear whether aeroaero-bic exercise in-creases or dein-creases volume in the hippocampus.
Volumetric changes of the hippocampus are thought to be elicited by neurogene-sis. Previous studies revealed that changes in neurogenesis can be associated with chan-ges in hippocampal volume. For example Biedermann et al. (2016) found a positive cor-relation between exercise induced grey matter volume increase and increased hippocam-pal neurogenesis in mice. Also, Marlatt et al. (2012) found elevated BDNF levels and neurogenesis after exercise in mice. Furthermore BDNF was closely linked to exercise induced hippocampal volume change in humans (Erickson et al., 2011; Wagner et al., 2015). Thus, BNDF levels seem to be associated with both hippocampal volume change and neurogenesis, contributing to the link between hippocampal volume change and neurogenesis.
Interestingly, the type of exercise performed seems to play an important role in the structural changes in the hippocampus. High intensity exercise induces more changes in fitness (VO2max) than low intensity exercise (Gormley et al. 2008; Helgerud et al.,
2007). For example, high intensity exercise leads to more changes in the neurogenesis mediator BDNF and bilateral hippocampal volume than low intensity exercise (Erickson et al., 2011). Also Hötting et al. (2016) found that only high intensity exercise increased BDNF levels, whereas low intensity exercise or relaxation did not. A similar result was
found by Nokia at al. (2016). Aerobic exercise lead to neurogenesis, but no neurogenesis was found in rats of the anaerobic exercise group. Thus, high intensity exercise seems to increase fitness levels (VO2max), hippocampal volume and neurogenesis (BDNF) more
than low intensity exercise. It therefore seems to matter how physical exercise is conduc-ted and at which intensity level in order to elicit changes in fitness and structural changes in the brain.
This study therefore tries to answer whether there is an effect of physical exercise on neurogenesis and volume in the healthy young adult human hippocampus. Based on the literature described above we expect that changes in cardiorespiratory fitness (VO2
-max) induced by physical exercise positively predict neurogenesis and therefore BDNF levels in humans (Marlatt et al., 2012). Furthermore we expect physical exercise induced changes in cardiorespiratory fitness (VO2max) to positively predict changes in
hippo-campal volume. (Erickson et al., 2011; Maass et al., 2016). Additionally, in line with pre-vious studies, we expect neurogenesis (BDNF levels) to positively predict volume change in the human hippocampus (Biedermann et al., 2016). And last we expect that high in-tensity exercise will improve cardiorespiratory fitness more and result in more changes in neurogenesis (BDNF levels) and hippocampal volume compared to low intensity exer-cise (Chaddock et al., 2010; Erickson et al., 2009; Griffin et al., 2011).
3. Materials and Methods
3.1 Participants and Experimental Design
In a controlled 12 week intervention, 52 sedentary healthy young adults (30 females, mean age = 24.43 years, SD = 2.59; 22 males, mean age = 22.59 years, SD= 3.08) signed written informed consent. The study was approved by the Medical Research Ethics Committee (MREC). Inclusion criteria were: age between 18-30 years, BMI ≤ 30 kg/m2,
VO2max≤ 55 ml/kg/min (males), VO2max≤ 45 ml/kg/min (females), use of oral
con-traceptive or intrauterine device (females) and a stable exercise history 3 months prior to study inclusion. Subjects who met any of the following criteria were excluded from par-ticipation in the study: general contraindications for MRI (claustrophobia), history of
psychiatric disorders, excessive smoking (> pack/day), excessive alcohol consumption (> 21 units/ week), or other regular drug use. Furthermore, participants were excluded if they engaged in intensive sports prior to the study intervention (>3 times/week). Sub-jects received monetary compensation after complete participation in the study.
3.2 Intervention and Procedure
3.2.1 Assessment of Fitness Levels (VO₂max, HR)
Subjects underwent a cardiopulmonary exercise test (volume oxygen maximum: VO2
-max) prior to and after the intervention, which assesses aerobic/cardiorespiratory fit-ness. Oxygen uptake by the tissue/VO2max is measured in mL/kg/min (Astorino &
Schubert, 2014). A recumbent cycle ergometer was used for the physical examination during which a pulmonary gas exchange was continuously obtained. Prior to the first test, the system was calibrated to a known concentration and to room air. Caffeine con-sumption, height, weight, fat percentage and waist size were assessed prior to testing. Heart rate was measured with a heart rate monitor and oxygen uptake was measured with an oxygen mask attached to mouth and nose. The first two minutes contained rest measurements of heart rate and oxygen up take during which participants were not cy-cling or talking. During the testing phase, the intensity level remained constant for three minutes and increased every minute thereafter until maximum capacity was reached and a two minute recovery phase followed.
3.2.2 Exercise Intervention
Subjects were randomly assigned either to the aerobic high intensity exercise group (n = 24) or the stretching and toning low intensity exercise group (n = 23). Participants of the aerobic exercise group followed a twelve-week exercise program in which they engaged in an aerobic exercise session for at least 45 min per session 3 times/week (Haskell et al., 2007; Astorino et al., 2014). At least two of the weekly sessions were interval exercise. Participants were told to engage in high intensity exercise, but were free to choose from any exercise courses at their local gym. High intensity exercise was distinguished from low intensity exercise by heart rate zones (0-100%). Participants of the high intensity
ex-ercise group were asked to train in heart rate zones 4-5 (60-100%). Participants of the low intensity exercise group were asked to train in heart rate zones 3 or lower. Further-more participants wore a Polar device around their chest in order to assess heart rate of each training session. Intensity level (heart rate) and duration of each training session were tracked online (https://www.polar.com/nl/producten/accessoires/h10_hartslag-sensor).
3.2.3 Measurement of Growth Factor Levels
Blood samples were obtained at baseline and post-exercise scan. Blood was drawn from a IV cannula and 8 mL blood samples were collected in order to analyze neuronal growth factors. The samples were centrifuged and serum was stored at -80° Celsius. These blood samples were used to assess levels of BDNF. BDNF was quantified using quantitative sandwich enzyme immunoassay technique ELISAs (https://www.rndsys-tems.com/), following the protocol of the manufacturer (R&D Systems, DBD00 for BDNF). Results were calculated according to the R&D protocol. Control (baseline) con-centrations, samples and their duplicates were averaged. Then, the control (baseline) concentration average was subtracted from the mean of each sample duplicate, of the pre- and post exercise samples respectively. According to the R&D protocol, a standard curve was created by plotting the mean absorbance for each standard optical density (OD) on the (y-axis) against the concentration on the (x-axis). A best fit curve was drawn through the points on the graph (regression line). The data was linearized by plot-ting the log of the human BNDF concentration against the log of the OD. The best fit was determined by regression analysis. The formulas of the regression analysis of the two plates were y1=0.16345x-0.2477 and y2=0.17365x-0.2655, where y refers to the O.D and x refers to the slope of the regression of the human free BDNF concentration. The formulas were solved for x (human free BDNF): x1 = (y1+0.2477)/(0.16345); x2 = (y2+0.2655)/(0.17365). Thus y1 and y2 (OD) were filled in with the values that were ob-tained by subtracting the control (baseline) average from the averages of each sample duplicate. The resulting concentrations were in pg/mL.
3.2.4 Magnetic Resonance Imaging: 7 Tesla high-resolution MRI
High resolution structural T1-weighted images and T2-weighted images were acquired within one week prior to and after the exercise intervention to capture pre- vs. post-in-tervention changes using a 7T MR system (Philips, Best, the Netherlands; with a 32 channel head coil). The parameters of sequences were the following. Three dimensional (3D) sagittal T1-weighted magnetisation images: repetition time (TR) = 4.115 ms; echo time (TE) = 1.847 ms; flip angle = 7 °; voxel dimensions = 1.09 x 0.89 x 0.67 mm; slice thickness = 0.9mm with no gap, 200 slices, matrix of 200 x 268 x 266; 240 x 180 mm field of view (FOV); acquisition time = 5.06 min. High resolution T2-weighted images were obtained as follows. TR = 6000 ms; TE = 80 ms; flip angle = 110°; voxel dimen-sions = 2 x 0.4 x 0.33 mm; slice thickness = 2.0 with no gap, 30 slices, matrix of 30 x 600 x 551; 240 x 180 mm FOV; acquisition time = 6.8 min.
3.2.5 Automated Segmentation 3.2.5.1 Freesurfer (version 6.0)
The segmentation of the hippocampus was performed with Freesurfer (version 6.0) (http://surfer.nmr.mgh.harvard.edu/). In brief, this technique estimates the probability of each voxel belonging to a certain brain region. This is done with a probabilistic atlas and a Bayesian modeling approach (Iglesias et al., 2015; Van Leemput et al., 2009; Iglesi-as, Sabuncu, & Van Leemput, 2013). The Freesurfer atlas assigns different voxels to dif-ferent colors, which are a linear combination of the subregion color codes, weighted by their corresponding probabilities (Iglesias et al., 2015, see Fischl et al., 2002 for a com-plete description of Freesurfer processing stages). The segmentation volumes were used for the analysis. Thirteen regions of the hippocampus were segmented and calculated, including CA1, CA2/3, CA4, (granule cell layer of) DG, fimbria, presubiculum, subicu-lum, parasubicusubicu-lum, molecular layer, hippocampus-amygdala-transition-area, hippocam-pal tail, whole hippocampus and hippocamhippocam-pal fissure (http://www.freesurfer.net/fswi-ki/HippocampalSubfields). Intensity bias corrected images (done in SPM12) were used as input for the longitudinal hippocampal pipeline. Longitudinal volume measures were
done in Freesurfer 6.0, with the T1-weighted images, calculating volume differences per subject making use of the pre- and post-intervention scans.
3.2.5.2 Automatic Segmentation of Hippocampal Subfields (ASHS)
For the comparison to the Freesurfer Segmentation, the open-source Automatic Seg-mentation of Hippocampal Subfields (ASHS) software (https://sites.google.com/site/ hipposubfields/) was used (Yushkevich et al., 2015). Therefore the latest version of ASHS (ashs-fastashs-20170223) and its corresponding atlas (UPenn PMC Atlas 20161128) were downloaded. The resulting segmentations and its colors correspond to the label with the highest probability at each subfield region (Yushkevich et al., 2015). As in the Freesurfer segmentation, the input images were T1- and T2-weighted images of each subject. Compared to Freesurfer, ASHS has no longitudinal pipeline. ASHS was therefore only used for a baseline segmentation comparison with Freesurfer (see Ap-pendix). ASHS is based on a training pipeline and a segmentation pipeline. In short, a multi-atlas joint label fusion and voxel-wise learning-based error correction (Wang et al., 2011; Wang et al., 2013; Yushkevich et al., 2015). Volumetric subfield calculations were performed fully automatically by the software and took approximately two hours per subject. The corr_usegray output was used for statistical analysis. Ten volumes of hip-pocampal subregions resulted from the analysis: CA1, CA2, CA3, DG, subiculum, ent-orhinal cortex, Brodmann area 35, Brodmann area 36, collateral sulcus and miscel-laneous parts.
3.2.5.3 Statistical Parametric Mapping: SPM 12
Intracranial volume (ICV) consists of grey matter, white matter and central spinal fluid (CSF) and was derived from SPM12 (http://www.fil.ion.ucl.ac.uk/spm/) with image segmentation. ICV was used as a control for Freesurfer and ASHS ICV outputs (in mm3). ICV was calculated for each subject. Because individuals vary in overall head size,
which leads to variations in volumes of interest (VOI), e.g. hippocampus. The VOIs the-refore need to be adjusted with each individual ICV accordingly. However, the relations-hip between VOI and ICV is neither proportional, nor linear but follows the power law
principle and can be expressed as VOI=𝛼*ICV𝛽 (Liu et al., 2014). Therefore 𝛽 was esti-mated with a non-linear regression analysis, with the VOI as the response variable and the ICV as the only covariate. The resulting 𝛽 estimates of each VOI were then used to correct each VOI with the ICV. Therefore the power proportion correction of ICV was used with the following formula: VOIPPC=VOI/ICV𝛽 (Liu et al., 2014). The resulting
power proportion corrected VOIs were then used to calculate the volumetric changes over time (Δ VOIPPC = VOIPPC post - VOIPPC pre) for each subject and of each
hemis-phere.
3.3 Statistical Analysis
Statistical analysis was conducted with SPSS for windows, version 22.0 (SPSS Inc. Chica-go, Illinois, USA). An independent samples t-test or Chi-squared test was used to assess baseline differences between the high and low exercise intensity groups for continuous and categorical variables respectively (Table 1). A paired samples t-test was done (with 95% confidence interval) in order to assess differences in the experimental groups over time (pre- and posttest). This was done in order to gain insight as to whether there are other unknown factors concurrently affecting the active manipulation besides the known manipulation (physical exercise training). Furthermore repeated measures ANOVA was conducted with VO2max pre (kg/ml/min), VO2max post (kg/ml/min) as within
sub-jects variable, group as between subsub-jects variable, gender and age as covariates. This was necessary in order to compare the cardiorespiratory fitness of high and low intensity ex-ercise groups before and after the intervention.
A linear regression analysis was conducted to assess the effect of physical exercise on neurogenesis (BDNF) and hippocampal volume (volume derived from T1-weighted images and IGF-1). The data was checked for potential bias in each regression analysis. Normality was assessed with Kolmogorov-Smirnov, Shapiro-Wilk-test and with P-P plots. In the case that the assumption of normality was violated, bootstrapping was per-formed with a 95% confidence interval. Linearity and homoscedasticity were checked with Levene's test and by plotting the regression standardised residual against the
regres-sion standardised predicted value. The assumption of independent errors was tested with Durbin Watson. Outliers have been checked with a partial plot, with z-scores and with boxplot. Extreme cases were inspected with Cook's distance, average leverage, Ma-halanobis distance, DFBeta and covariance ratio. The fourth hypothesis contains two predictors. Therefore, multicollinearity was checked with Variance Inflation Factor (VIF).
Hippocampal volumes were derived from the segmentation performed with Free-surfer (version 6.o) (https://surfer.nmr.mgh.harvard.edu/). The resulting volumes of the hippocampus and the corresponding subfields were corrected with ICV for each hemisphere. The influence of exercise intensities and brain volume was measured with regression analyses (Lüders, Steinmeggtz & Jäncke, 2002). First, the change in cardiore-spiratory fitness (ΔVO2max) was the independent variable and the change in
neurogene-sis (ΔBDNF) was the dependent variable (Hypotheneurogene-sis 1). Second, the change in cardio-respiratory fitness (ΔVO2max) was the independent variable and the change in
hippo-campal volume (ΔVOIPPC) was the dependent variable (Hypothesis 2). Third, the change
in neurogenesis (ΔBDNF) was the in dependent variable and hippocampal volume change (ΔVOIPPC) was the dependent variable (Hypothesis 3). And last, the change in
cardiorespiratory fitness (ΔVO2max) was the independent variable and the change in
hippocampal volume (ΔVOIPPC) and neurogenesis (ΔBDNF) were the dependent
varia-bles, with group as a regressor (Hypothesis 4). It is worthy to note that the regression analysis of VOI differences over time (ΔVOIPPC) were limited to the whole
hippocam-pus and the DG of the left and right hemisphere and were not conducted over all hip-pocampal subfields. Detailed subfield volumetric data can be found in the Appendix (Table 3).
4. Results
a Independent samples t-test, b Chi-squared test, * indicates significance at p<.05.
Table 1 Participants’ Characteristics and Exercise Information
High intensity Exer-cise Group (N = 24)
Mean (SD)
Low intensity Exer-cise Group (N= 23) Mean (SD) p-value Sex (male/female) 10/14 11/12 0.671b Age (years) 22.79 (2.32) 24.04 (3.21) 0.131a Weight baseline (kg) 71.08 (12.82) 72.16 (10.46) 0.754a Weight post (kg) 70.91 (13.04) 72.66 (10.17) 0.614a ΔWeight (kg) -0.16 (1.98) -0.50 (2.36) 0.302a
VO2max (kg/ml/min) baseline 37.14 (7.33) 36.79 (5.81) 0.855a
VO2max (kg/ml/min) baseline 39.24 (7.16) 37.92 (7.54) 0.542a
ΔVO2max (kg/ml/min) 2.10 (5.18) 1.14 (5.32) 0.533a
BDNF (pg/mL) baseline 4.62 (0.92) 4.97 (1.04) 0.224a
BDNF (pg/mL) post 4.80 (0.73) 4.79 (0.99) 0.507a
ΔBDNF (pg/mL) 0.24 (1.12) -0.003 (1.10) 0.475a
Total time spent exercising (h) 30.40 (15.75) 29.62 (11.84) 0.848a
Total recorded exercise time
polar (h) 18.03 (10.46) 20.89 (9.14) 0.325
a
Average training duration per
session (h) 0.85 (0.26) 0.91 (0.18) 0.363
a
Total number of trainings 33.54 (12.02) 31.74 (9.01) 0.565a
Time of HR above 75% (h) 7.90 (4.62) 2.77 (1.77) 0.000a * Fraction of HR time above
75% of total recording time 47.51 (16.16) 12.59 (6.42) 0.000 a *
Intelligence (DART) 85.30 (6.66) 84.41 (8.31) 0.691a
4.1. Demographics and Manipulation Check
A priori sample size calculation of 52 subjects was based upon a moderate effect size for neurogenesis in humans of f2=0.16. To detect the effect of exercise, with a power of
80% and a α of 0.05, 19 subjects in the exercise intervention group were needed. Taking a 20% dropout into account (due to MRI data quality), at least 24 subjects were required for the exercise group (sample size calculation were done with G*Power (Faul, Erdfel-der, Lang & Buchner, 2007) (http://www.gpower.hhu.de/en.html).
In total, 52 subjects were randomly assigned to either the high intensity exercise group (n=24) or the low intensity group (n=23). There were five dropouts prior or du-ring the intervention. In total, 47 subjects participated in the study (26 females, 21 males, mean age = 23.40, SD = 2.83 years). Groups did not differ with regard to sex (Chi-squa-red), age, weight, cardiorespiratory fitness (VO2max), BDNF levels, total time spent
ex-ercising, total recorded exercise time (Polar), average training duration, total number of trainings, intelligence and years of education (independent samples t-test, Table 1). The high intensity group had spent more time with heart rate above 75% (high intensity, 7.90 (4.62); low intensity, 2.77 (1.77), p < 0.05 by t-test) and had a higher proportion of heart rate time above 75% (high intensity, 47.51 (16.16); low intensity 12.59 (6.42), p < 0.05 by
t-test) compared to the low intensity group (Table 1). A paired samples t-test was
per-formed in order to assess whether there are other unknown factors concurrently affec-ting the active manipulation besides the known manipulation (physical exercise). Average cardiorespiratory fitness was better after the physical exercise intervention (VO2max; M
= 38.97, SE = 6.56) compared to before the intervention (M = 36.60, SE = 7.30). This difference was significant t(46) = -2.140, p = 0.038, and presented a small effect size, dCohen = 0.234. See Table 4 (Appendix) for a comparison of hippocampal volumetric
4.2 Cardiorespiratory Fitness and Neurogenesis
A multiple linear regression was used to assess whether the change in cardiorespiratory fitness (ΔVO2max) predicts change in neurogenesis (ΔBDNF). Due to the violated
as-sumption of linearity and normality, bootstrapping was performed with a 95% confi-dence interval. The change in cardiorespiratory fitness (ΔVO2max), b = 0.007 [-0.08,
0.07], p = 0.868, did not significantly predict change in neurogenesis (ΔBDNF). 4.3 Cardiorespiratory Fitness and Hippocampal Volume
In order to test whether the change in cardiorespiratory fitness (ΔVO2max) predicts the
change in hippocampal volume (ΔVOIPPC) we conducted a multiple linear regression
us-ing the whole hippocampus and the DG for each hemisphere. The distribution of the volumetric data of the left whole hippocampus was not normally distributed. Therefore, bootstrapping with a 95% confidence interval was performed for the first regression (see
Δ H ip po ca mp us le ft he mis ph er e ( mm ³) -6 -2,25 1,5 5,25 9 ΔVO2max (kg/ml/min) -12 -4,75 2,5 9,75 17 y = -0,089x - 1,0115 R² = 0,0406 Δ H ip po ca mp us r ig ht he mis ph er e ( mm ³) -9,00 -5,00 -1,00 3,00 7,00 ΔVO2max (kg/ml/min) -12 -4,75 2,5 9,75 17 y = -0,0256x - 1,3941 R² = 0,0046 Δ DG le ft h emis ph er e (mm ³) -1,4 -0,3 0,8 1,9 3 ΔVO2max (kg/ml/min) -12 -4,75 2,5 9,75 17 y = -0,027x + 0,0279 R² = 0,0307 Δ DG r ig ht h emis ph er e (mm ³) -3 -1,375 0,25 1,875 3,5 ΔVO2max (kg/ml/min) -12 -4,75 2,5 9,75 17 y = -0,0013x - 0,6506 R² = 0,0001
Fig.1. The relationship between the change in cardiorespiratory fitness (ΔVO2max) and
the change in volume of A) the left hippocampus (ΔHippocampus left); B) the right hippocampus (ΔHippocampus right); C) the left DG (ΔDG left) and D) the right DG
A) B)
Figure 1). The change in cardiorespiratory fitness (ΔVO2max), b = -0.09 [-0.21, 0.04], p
= .196, did not significantly predict the change in hippocampal volume of the left hemi-sphere (ΔHippocampus left). The change in cardiorespiratory fitness (ΔVO2max), t(45)
= -0.457, p = .650, did not significantly predict the change in hippocampal volume of the right hemisphere (ΔHippocampus right). Furthermore, the change in cardiorespira-tory fitness (ΔVO2max), t(45) = -1.19, p = .239, did not significantly predict the change
in DG volume of the left hemisphere (ΔDG left). And lastly, the change in cardiorespi-ratory fitness (ΔVO2max), t(45) = -0.069, p = .945, did not significantly predict the
change in DG volume of the right hemisphere (ΔDG right). 4.4 Neurogenesis and Hippocampal Volume
A linear regression was used to test whether the change in BDNF (ΔBDNF) predicts the change in hippocampal volume (ΔVOIPPC). Again, the whole hippocampus and the DG
Δ H ip po ca mp us le ft he mis ph er e ( mm ³) -6 -2,25 1,5 5,25 9 ΔBDNF (pg/mL) -2 -1 0 1 2 3 4 y = -0,3325x - 1,2493 R² = 0,0256 Δ H ip po ca mp us r ig ht he mis ph er e ( mm ³) -9,00 -4,88 -0,75 3,38 7,50 ΔBDNF (pg/mL) -2 -1 0 1 2 3 4 y = -0,0558x - 1,4078 R² = 0,001 Δ DG le ft h emis ph er e (mm ³) -1,4 -0,05 1,3 2,65 4 ΔBDNF (pg/mL) -2 -1 0 1 2 3 4 y = -0,1203x - 0,0115 R² = 0,0265 Δ DG r ig ht h emis ph er e (mm ³) -3 -1,25 0,5 2,25 4 ΔBDNF (pg/mL) -2 -1 0 1 2 3 4 y = -0,173x - 0,6415 R² = 0,0444
Fig.2. The relationship between the change in BDNF (ΔBDNF) and the change in
volume of A) the left hippocampus (ΔHippocampus left); B) the right hippocampus (ΔHippocampus right); C) the left DG (ΔDG left) and D) the right DG (ΔDG right)*.
A) B)
of each hemisphere were entered in the regression. As stated above, the volumetric data of the change in hippocampus of the left hemisphere was not normally distributed. Nei-ther was the change in BDNF. Therefore bootstrapping was performed with a 95% con-fidence interval. The change in BDNF (ΔBDNF), b = -0.33 [-0.89, 0.23], p = .147, did not significantly predict the change in hippocampal volume of the left hemisphere (ΔHippocampus left). Furthermore the change in BDNF (ΔBDNF), b = -0.06 [-0.56, 0.47], p = .778 did not significantly predict the change in hippocampal volume of the right hemisphere (ΔHippocampus right). And the change in BDNF (ΔBDNF), b = -0.12 [-0.25, 0.03], p = .089 did not significantly predict the change in DG volume of the left hemisphere (ΔDG left). Lastly and interestingly, the change in BDNF (ΔBDNF), b = -0.173 [-0.35, -0.01], p = .041*, did significantly predict the change in DG volume of the right hemisphere (Figure 2).
4.5 Fitness Intensities, Neurogenesis and Hippocampal Volume
A linear regression was used to test whether the change in cardiorespiratory (ΔVO2max)
and exercise intensity (high vs low intensity, group was added as a regressor) predict the change in hippocampal volume (ΔVOIPPC). As stated above, change in hippocampus left
was not normally distributed. Therefore, first regression was performed with bootstrap-ping and a 95% confidence interval (Figure 3).
Δ H ip po ca mp us le ft h emis ph er e ( mm ³) -7 -5,25 -3,5 -1,75 0 1,75 3,5 5,25 7 ΔVO2max (kg/ml/min) -12 -8 -4 0 4 8 12 16 R² = 0,0057 R² = 0,0031
Hippocampus left, high intensity High intensity Hippocampus left, low intensity Low intensity
Δ H ip po ca mp us r ig ht h emis ph er e ( mm ³) -9,00 -6,75 -4,50 -2,25 0,00 2,25 4,50 -12 -8 -4 0 4 8 12 16 R² = 0,0012 R² = 0,0078
Hippocampus right, high intensity High intensity Hippocampus right, low intensity Low intensity
A)
The change in cardiorespiratory fitness (ΔVO2max), b = -0.09 [-0.23, 0.06], p = .206, and
exercise intensity, b = 0.05 [-1.30, 1.22], p = .935, did not significantly predict the change in hippocampal volume of the left hemisphere (ΔHippocampus left). The change in cardiorespiratory fitness (ΔVO max), t(44) = -0.47, p = .640, and exercise intensity, t(44)
Δ DG le ft h emis ph er e ( mm ³) -1,5 -1 -0,5 0 0,5 1 1,5 2 ΔVO2max (kg/ml/min) -12 -8 -4 0 4 8 12 16 R² = 0,0347 R² = 0,0539
DG left, high intensity DG left, high intensity DG left, low intensity DG left, low intensity
Δ DG r ig ht h emis ph er e ( mm ³) -3 -2,25 -1,5 -0,75 0 0,75 1,5 2,25 ΔVO2max (kg/ml/min) -12 -8 -4 0 4 8 12 16 R² = 0,0054 R² = 0,0066
DG right, high intensity DG right, high intensity DG right, low intensity DG right, low intensity
Fig.3. Relationship between the change in cardiorespiratory fitness (ΔVO2max),
exercise intensity and the change in volume of A) the left hippocampus (ΔHippocampus left); B) the right hippocampus (ΔHippocampus right); C) the left DG (ΔDG left) and D) the right DG (ΔDG right). Volumes are in cubic millimeters.
C)
= -0.23, p = .821, did not significantly predict the change in hippocampal volume of the right hemisphere (ΔHippocampus right). The change in cardiorespiratory fitness (ΔVO2max), t(44) = -1.19, p = .241, and exercise intensity, t (44) = -0.14, p = .887, did
not significantly predict the change in DG volume of the left hemisphere (ΔDG left). And lastly, the change in cardiorespiratory fitness (ΔVO2max), t(44) = -0.08, p = .941,
and exercise intensity, t(44) = -0.07, p = .943, did not significantly predict the change in DG volume of the right hemisphere (ΔDG right).
Furthermore, we wanted to test whether the change in cardiorespiratory (ΔVO2
-max) and exercise intensities (high vs low intensity, group was added as a regressor) pre-dict the change in neurogenesis (ΔBDNF). Again, this regression was bootstrapped with a 95% confidence interval (Figure 4). The change in cardiorespiratory fitness ΔVO2
-max, b = -0.004 [-0.08, 0.06], p = .913, and exercise intensity, b = -.24 [-0.89, 0.38], p = . 478, did not significantly predict ΔBDNF.
Δ B DNF (p g/mL ) -2 -1 0 1 2 3 4 ΔVO2max (kg/ml/min) -12 -8 -4 0 4 8 12 16 R² = 0,0019 R² = 0,0176
BDNF, high intensity BDNF, high intensity BDNF, low intensity BDNF, low intensity
Fig.4. The relationship between the change in cardiorespiratory fitness (ΔVO2max),
5. Discussion
There is converging evidence from animal models that physical exercise increases neu-rotrophic factors which are known to modulate hippocampal plasticity. However, exer-cise related neurotrophic and volumetric changes in the human brain have not been straightforward. Moreover, it is unclear whether exercise intensities can lead to a differ-ent degree in neurotrophic and volumetric changes. In this study we analyzed in healthy young adults whether physical exercise, at high and low intensity, has an effect on hip-pocampal volume and neurotrophic factor BDNF (measured before and after the inter-vention). Our data did not demonstrate that aerobic fitness leads to increased levels of the neurogenesis mediator BDNF, or increased hippocampal volume (Erickson et al., 2011). As did Wagner et al. (2015), we found a significant relationship between BDNF increase and a volumetric decrease of the right DG. This is unexpected, since BDNF and hippocampal volume were reported to correlate positively (Biedermann et al., 2016; Erickson et al., 2011). Furthermore, we were unable to confirm that higher levels of fit-ness lead to more changes in hippocampal volume and BDNF compared to low intensi-ty exercise (Erickson et al., 2009; Erickson et al., 2011; Hötting et al., 2016).
Contrary to our expectations, all results between cardiorespiratory fitness and hippocampal volume were non-significant. It is worthy to note that volumetric changes in the hippocampus seem to vary widely. Although, Erickson et al. (2011) did indeed find an overall positive relationship between fitness and hippocampal volume increase, their data demonstrated a wide variety of volumetric changes. They not only found an increase in hippocampal volume but also decrease in volume. For these reasons, structu-ral changes should be closely analyzed on an individual basis and an ovestructu-rall trend towards hippocampal increase should not be viewed as the norm. Furthermore, as Wag-ner et al. (2015), we examined a young population and also found a small hippocampal decrease after regular physical exercise. Most studies in the elderly found an exercise in-duced hippocampal volume increase. Thus, age may play an important role in exercise induced structural changes and is worth further investigation.
Above that, it is worthy to note that the hippocampal volumes derived from au-tomated segmentation can differ between segmentation techniques. Especially
volu-metric data of hippocampal subfields, differ between automated segmentation techni-ques such as Freesurfer v.6.0 and ASHS. We are aware of the fact that both segmentati-on techniques use different atlases. However, ASHS might be a more accurate technique. That is, the Freesurfer atlas was built from manual delineations in elderly subjects only, and might therefore contain hippocampal atrophy which makes it potentially less app-licable for segmenting brain scans of a younger population (Iglesias et al., 2015). Fur-thermore, the Freesurfer applicability has only been validated on lower field strengths (T1-weighted and T2-weighted MRI data up to 3 Tesla), but we obtained our data in a 7Tesla scanner (Guiliano et al., 2017). Therefore volumetric data should be interpreted with caution. Nevertheless, we did not conduct the main hippocampal subfield segmen-tation in ASHS, since it does not contain a longitudinal pipeline (for a comparison see Appendix, Table 3, Figure 5 and 6).
BDNF seems to play an important role in exercise related hippocampal volume change. But it is not clear whether BDNF/neurogenesis alone are the driving factor for hippocampal volume increase. Pereira et al. (2007) found a significant increase in Cereb-ral Blood Volume (CBV) in the DG after three months of regular aerobic exercise. Inte-restingly environmental enrichment, thus novel experiences seem to elicit hippocampal volume increase in mice (Scholz, Allemang-Grand, Dazai, & Lerch, 2015). Furthermore, exercise alone vs. exercise with environmental enrichment elicit different structural changes in the rodent brain: more new neurons are produced in the rodent hippocampus with only exercising, but exercise in combination with environmental enrichment leads to a greater survival of new neurons (Van Praag, Kempermann & Gage, 2000; Van Praag, Shubert, Zhao & Gage, 2005). Since humans are much more exposed to envi-ronmental enrichment than laboratory rodents, it is difficult to show this effect in a hu-man study. However, we cannot exclude the possibility that new social interactions, ta-king part in a scientific study and attending new physical exercise classes forms a poten-tial environmental enrichment that could contribute to hippocampal volume change.
One possible explanation for the non-significant results between cardiorespiratory fitness, volume, BDNF and exercise intensities might be an insufficient effect of heart rate (HR) zones on cardiorespiratory fitness (VO2max). We expected that spending more
time in HR zone above 75%, thus high intensity exercise, would elicit more changes in cardiorespiratory fitness (VO2max) compared to low intensity exercise (Erickson et al.,
2011). This was not the case in our study. We therefore concluded that the link between exercise intensities (heart rate zones) and cardiorespiratory fitness (VO2max) is not as
straightforward as previously thought. Although we tried to control training duration and frequency of training, effects of initial fitness levels can have a great impact on the improvement of cardiorespiratory fitness. Additionally, it is worth mentioning that there are so called non-responders to high intensity exercise, individuals who demonstrate litt-le to no improvement in fitness litt-levels (VO2max, exercise HR) as a result of genetic
fac-tors (Astorino et al., 2014). The question arising, is whether heart rate zones are the best possible way to distinguish exercise intensities, or whether VO2max/aerobic capacity
monitoring during exercising should be used instead, in order to guarantee that high in-tensity exercise really elicits aerobic changes. Another important aspect of this study is the lack of a physicaly inactive control group which makes it extremely difficult to gene-ralize the results.
Nevertheless, this study has shown that there is a strong link between BDNF and volumetric changes in the right DG, strengthening the theory that the DG is a unique birthplace of neurons. Our study also showed that physical exercise does impact the structure of the young human brain (slight shrinkage) differently than the elderly brain (increase in volume). This stresses that there are not only age related differences in struc-tural changes to physical exercise but there might also be great individual differences left to reveal. Only when we fully understand which individual factors contribute to success-ful exercise induced structural changes, will it be possible to tailor physical training in order to achieve the maximum benefit for cognition and mental health.
6. Literature
Åberg, M. A., Pedersen, N. L., Torén, K., Svartengren, M., Bäckstrand, B., Johnsson, T., ... & Kuhn, H. G. (2009). Cardiovascular fitness is associated with cognition in young adulthood. Proceedings of the National Academy of Sciences, 106(49), 20906-20911. de Almeida, A. A., da Silva, S. G., Fernandes, J., Peixinho-Pena, L. F., Scorza, F. A.,
Cav-alheiro, E. A., & Arida, R. M. (2013). Differential effects of exercise intensities in hip-pocampal BDNF, inflammatory cytokines and cell proliferation in rats during the postnatal brain development. Neuroscience Letters, 553, 1-6.
Anderson-Hanley, C., Arciero, P. J., Brickman, A. M., Nimon, J. P., Okuma, N., Westen, S. C., ... & Zimmerman, E. A. (2012). Exergaming and older adult cognition: a cluster randomized clinical trial. American Journal of Preventive Medicine, 42(2), 109-119.
Astorino, T. A., & Schubert, M. M. (2014). Individual responses to completion of short-term and chronic interval training: a retrospective study. PloS one, 9(5), e97638.
Biedermann, S. V., Fuss, J., Steinle, J., Auer, M. K., Dormann, C., Falfán-Melgoza, C., ... & Weber-Fahr, W. (2016). The hippocampus and exercise: histological correlates of MR-detected volume changes. Brain Structure and Function, 221(3), 1353-1363.
Boyke, J., Driemeyer, J., Gaser, C., Büchel, C., & May, A. (2008). Training-induced brain structure changes in the elderly. Journal of Neuroscience, 28(28), 7031-7035.
Buchman, A. S., Boyle, P. A., Yu, L., Shah, R. C., Wilson, R. S., & Bennett, D. A. (2012). Total daily physical activity and the risk of AD and cognitive decline in older adults.
Neurology, 78(17), 1323-1329.
Chaddock, L., Erickson, K. I., Prakash, R. S., VanPatter, M., Voss, M. W., Pontifex, M. B., ... & Kramer, A. F. (2010). Basal ganglia volume is associated with aerobic fitness in preadolescent children. Developmental Neuroscience, 32(3), 249-256.
Duchesne, C., Gheysen, F., Boré, A., Albouy, G., Nadeau, A., Robillard, M. E., ... & Do-yon, J. (2016). Influence of aerobic exercise training on the neural correlates of motor learning in Parkinson's disease individuals. NeuroImage: Clinical, 12, 559-569.
Erickson, K. I., Prakash, R. S., Voss, M. W., Chaddock, L., Hu, L., Morris, K. S., ... & Kramer, A. F. (2009). Aerobic fitness is associated with hippocampal volume in elderly humans. Hippocampus, 19(10), 1030-1039.
Erickson, K. I., Voss, M. W., Prakash, R. S., Basak, C., Szabo, A., Chaddock, L., ... & Wo-jcicki, T. R. (2011). Exercise training increases size of hippocampus and improves memory. Proceedings of the National Academy of Sciences, 108(7), 3017-3022.
Erickson, K. I., Miller, D. L., & Roecklein, K. A. (2012). The aging hippocampus: inter-actions between exercise, depression, and BDNF. The Neuroscientist, 18(1), 82-97. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible
statisti-cal power analysis program for the social, behavioral, and biomedistatisti-cal sciences. Behavior
Research Methods, 39, 175-191.
Fischl, B., Salat, D. H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., ... & Montillo, A. (2002). Whole brain segmentation: automated labeling of neuroanatomical struc-tures in the human brain. Neuron, 33(3), 341-355.
Giuliano, A., Donatelli, G., Cosottini, M., Tosetti, M., Retico, A., & Fantacci, M. E. (2017). Hippocampal subfields at ultra high field MRI: An overview of segmentation and measurement methods. Hippocampus, 27(5), 481-494.
Gormley, S. E., Swain, D. P., High, R. E. N. E. E., Spina, R. J., Dowling, E. A., Kotipalli, U. S., & Gandrakota, R. A. M. Y. A. (2008). Effect of Intensity of Aerobic Training on VO2 max. Medicine and Science in Sports and Exercise, 40(7), 1336.
Gottmann, K., Mittmann, T., & Lessmann, V. (2009). BDNF signaling in the formation, maturation and plasticity of glutamatergic and GABAergic synapses. Experimental
Brain Research, 199(3-4), 203-234.
Griffin, É. W., Mullally, S., Foley, C., Warmington, S. A., O'Mara, S. M., & Kelly, Á. M. (2011). Aerobic exercise improves hippocampal function and increases BDNF in the serum of young adult males. Physiology & Behavior, 104(5), 934-941.
Groot, C., Hooghiemstra, A. M., Raijmakers, P. G. H. M., van Berckel, B. N. M., Schel-tens, P., Scherder, E. J. A., ... & Ossenkoppele, R. (2016). The effect of physical activi-ty on cognitive function in patients with dementia: a meta-analysis of randomized control trials. Ageing Research Reviews, 25, 13-23.
Haskell, W. L., Lee, I. M., Pate, R. R., Powell, K. E., Blair, S. N., Franklin, B. A., ... & Bauman, A. (2007). Physical activity and public health: updated recommendation for
adults from the American College of Sports Medicine and the American Heart Asso-ciation. Circulation, 116(9), 1081.
Helgerud, J., Hoydal, K., Wang, E., Karlsen, T., Berg, P., Bjerkaas, M., ... & Hoff, J. (2007). Aerobic High-Intensity Intervals Improve VO~ 2~ m~ a~ x More Than Mo-derate Training. Medicine and Science in Sports and Exercise, 39(4), 665.
Hötting, K., Schickert, N., Kaiser, J., Röder, B., & Schmidt-Kassow, M. (2016). The Ef-fects of Acute Physical Exercise on Memory, Peripheral BDNF, and Cortisol in Young Adults. Neural Plasticity, 2016. https://doi.org/10.1155/2016/6860573
Iglesias, J. E., Sabuncu, M. R., Van Leemput, K., & Alzheimer’s Disease Neuroimaging Initiative. (2013). Improved inference in Bayesian segmentation using Monte Carlo sampling: Application to hippocampal subfield volumetry. Medical Image Analysis, 17(7), 766-778.
Iglesias, J. E., Augustinack, J. C., Nguyen, K., Player, C. M., Player, A., Wright, M., ... & Fischl, B. (2015). A computational atlas of the hippocampal formation using ex vivo, ultra-high resolution MRI: application to adaptive segmentation of in vivo MRI.
Neu-roimage, 115, 117-137.
Jonasson, L. S., Nyberg, L., Kramer, A. F., Lundquist, A., Riklund, K., & Boraxbekk, C. J. (2016). Aerobic exercise intervention, cognitive performance, and brain structure: Re-sults from the Physical Influences on Brain in Aging (PHIBRA) study. Frontiers in
Ag-ing Neuroscience, 8, 336.
Karege, F., Schwald, M., & Cisse, M. (2002). Postnatal developmental profile of brain-derived neurotrophic factor in rat brain and platelets. Neuroscience Letters, 328(3), 261-264.
Kempermann, G., Wiskott, L., & Gage, F. H. (2004). Functional significance of adult neurogenesis. Current Opinion in Neurobiology, 14(2), 186-191.
Kempermann, G. (2012). New neurons for'survival of the fittest'. Nature Reviews
Neuro-science, 13(10), 727-736.
Liu, D., Johnson, H. J., Long, J. D., Magnotta, V. A., & Paulsen, J. S. (2014). The power-proportion method for intracranial volume correction in volumetric imaging analysis.
Lüders, E., Steinmetz, H., & Jäncke, L. (2002). Brain size and grey matter volume in the healthy human brain. Neuroreport, 13(17), 2371-2374.
Maass, A., Düzel, S., Brigadski, T., Goerke, M., Becke, A., Sobieray, U., ... & Braun-Dul-laeus, R (2016). Relationships of peripheral IGF-1, VEGF and BDNF levels to exer-cise-related changes in memory, hippocampal perfusion and volumes in older adults.
Neuroimage, 131, 142-154.
Manganas, L. N., Zhang, X., Li, Y., Hazel, R. D., Smith, S. D., Wagshul, M. E., ... & Ma-letić-Savatić, M. (2007). Magnetic resonance spectroscopy identifies neural progenitor cells in the live human brain. Science, 318(5852), 980-985.
Marlatt, M. W., Potter, M. C., Lucassen, P. J., & van Praag, H. (2012). Running throughout middle‐age improves memory function, hippocampal neurogenesis, and BDNF levels in female C57BL/6J mice. Developmental Neurobiology, 72(6), 943-952. Nokia, M. S., Lensu, S., Ahtiainen, J. P., Johansson, P. P., Koch, L. G., Britton, S. L., &
Kainulainen, H. (2016). Physical exercise increases adult hippocampal neurogenesis in male rats provided it is aerobic and sustained. The Journal of Physiology, 594(7), 1855-1873.
Pereira, A. C., Huddleston, D. E., Brickman, A. M., Sosunov, A. A., Hen, R., McKhann, G. M., ... & Small, S. A. (2007). An in vivo correlate of exercise-induced neurogenesis in the adult dentate gyrus. Proceedings of the National Academy of Sciences, 104(13), 5638-5643.
Sahay, A., Scobie, K. N., Hill, A. S., O'carroll, C. M., Kheirbek, M. A., Burghardt, N. S., ... & Hen, R. (2011). Increasing adult hippocampal neurogenesis is sufficient to im-prove pattern separation. Nature, 472(7344), 466-470.
Schmand, B., Bakker, D., Saan, R., & Louman, J. (1991). The Dutch Reading Test for Adults: a measure of premorbid intelligence level. Tijdschrift voor Gerontologie en Geriatrie,
22(1), 15-19.
Scholz, J., Allemang-Grand, R., Dazai, J., & Lerch, J. P. (2015). Environmental enrich-ment is associated with rapid volumetric brain changes in adult mice. NeuroImage, 109, 190-198.
Spalding, K. L., Bergmann, O., Alkass, K., Bernard, S., Salehpour, M., Huttner, H. B., ... & Possnert, G. (2013). Dynamics of hippocampal neurogenesis in adult humans. Cell,
153(6), 1219-1227.
Van Leemput, K., Bakkour, A., Benner, T., Wiggins, G., Wald, L. L., Augustinack, J., ... & Fischl, B. (2009). Automated segmentation of hippocampal subfields from ultra‐high resolution in vivo MRI. Hippocampus, 19(6), 549-557.
Van Praag, H., Kempermann, G., & Gage, F. H. (2000). Neural consequences of envi-romental enrichment. Nature Reviews Neuroscience, 1(3), 191-198.
Van Praag, H., Shubert, T., Zhao, C., & Gage, F. H. (2005). Exercise enhances learning and hippocampal neurogenesis in aged mice. Journal of Neuroscience, 25(38), 8680-8685. Wagner, G., Herbsleb, M., Cruz, F. D. L., Schumann, A., Brünner, F., Schachtzabel, C., ...
& Reichenbach, J. R. (2015). Hippocampal structure, metabolism, and inflammatory response after a 6-week intense aerobic exercise in healthy young adults: a controlled trial. Journal of Cerebral Blood Flow & Metabolism, 35(10), 1570-1578.
Wang, H., Das, S. R., Suh, J. W., Altinay, M., Pluta, J., Craige, C., ... & Alzheimer's Disease Neuroimaging Initiative. (2011). A learning-based wrapper method to correct system-atic errors in automsystem-atic image segmentation: consistently improved performance in hippocampus, cortex and brain segmentation. NeuroImage, 55(3), 968-985.
Wang, H., Suh, J. W., Das, S. R., Pluta, J. B., Craige, C., & Yushkevich, P. A. (2013). Multi-atlas segmentation with joint label fusion. IEEE Transactions on Pattern Analysis and
Ma-chine Intelligence, 35(3), 611-623.
Wei, Z., Liao, J., Qi, F., Meng, Z., & Pan, S. (2015). Evidence for the contribution of BDNF-TrkB signal strength in neurogenesis: an organotypic study. Neuroscience Letters, 606, 48-52.
Yushkevich, P. A., Pluta, J. B., Wang, H., Xie, L., Ding, S. L., Gertje, E. C., ... & Wolk, D. A. (2015). Automated volumetry and regional thickness analysis of hippocampal sub-fields and medial temporal cortical structures in mild cognitive impairment. Human
7. Appendix
Table 2 Summary of the Anatomical Labels for Freesurfer (v.6.0) (Iglesias et al., 2015)
and ASHS (Yushkevich et al., 2015)
Abbr Name Comments
ASHS
CA1 Cornu ammonis field 1
CA2 Cornu ammonis field 2
CA3 Cornu ammonis field 3
CA Cornu ammonis field 1,2 and 3
DG Dentate gyrus
Sub Subiculum
HIPP Whole Hippocampus Volume of whole hippocampus is
deri-ved by the sum of CA, DG, Sub.
ERC Entorhinal Cortex
BA35 Brodmannarea 35
BA36 Brodmannarea 36
PRC Perirhinal Cortex Combines labels BA35 and BA36
Misc Miscellaneous
CS Collateral Sulcus
Freesurfer
PreSub Presubiculum Brodmann's area 27
CA1 Cornu ammonis field 1
CA23 Cornu ammonis field 2 and 3
combined
Fimb Fimbria White matter structure that extends
from alveus and forms the fornix fur-ther on.
CA4/DG Cornu ammonis field 4 Hilar region of denate gyrus. CA4 lies within the DG.
GC-DG Granule cell layer of dentate
gyrus DG consists of the molecular layer, granule cell layer and a polymorphic layer.
HIPP Whole Hippocampus Default output of Freesurfer segmenta-tion.
H. Fissure Hippocampal fissure
ParaSub Parasubiculum Brodmann’s area 49
ML Molecular layer Molecular layer of subiculum or CA
fields. Lies beneath hippocampal fissure or above the subiculum.
HATA
Hippocampal-amygdala-transi-tion-area Lies in the medial region of the hippo-campus. The HATA lies superior to the other subfields.
Tail Hippocampal tail.
Table 2 Summary of the Anatomical Labels for Freesurfer (v.6.0) (Iglesias et al., 2015)
and ASHS (Yushkevich et al., 2015)
Table 4 Mean (SD) and p-values of Raw Volumes, Power Proportion Corrected
(PPC), and Proportion Corrected Volumes between Exercise Intensities and Over Time (T0, T1)
Volumes High
In-tensity Baseline (N=24) High In-tensity Post (N=23) High In-tensity Change (T1-T0) Low Inten-sity Baseli-ne (N=24) Low Inten-sity Post (N=23) Low In-tensity Change (T1-T0) p-va-lue
Pre vs Post Whole Hippocampus un-corrected left 3782.69 (429.2) 3779.60 (401.7) -3.09 3613.99 (314.6) 3622.89 (315.9) 8.9 0.64 6
Pre vs Post Whole Hippocampus un-corrected right 3741.71 (915.69) 3735.26 (907.72) -6.45 3759.08 (347.15) 3755.06 (364.44) -4.02 0.910
Pre vs Post Whole Hippocampus PPC left 97.31 (19.43) 96.09 (20.44) -1.22 90.76 (16.72) 89.68 (16.24) -1.08 0.850
Pre vs Post Whole Hippocampus PPC right 92.32 (27.88) 90.94 ( 27.74) -1.38 90.51 (17.12) 89.02 (15.91) -1.49 0.853
Pre vs Post Whole Hippocampus Proportion cor-rected left 0.0041 (0.06) 0.0042 (0.006) 0.00005 0.0033 (0.0044) 0.0033 (0.0044) -0.000002 0.277
Pre vs Post Whole Hippocampus Proportion cor-rected right 0.0041 (0.006) 0.0041 (0.006) 0.0000038 0.0034 (0.0047) 0.0034 (0.0045) -0.000042 0.380 Pre vs Post DG uncorrected left (78.50)329.72 (78.37)329.33 -0.39 334.39 (29.16) (27.60)334.34 -0.05 0.919 Pre vs Post DG uncorrected right (84.94)346.62 (84.56)346.43 -0.19 351.69 (34.03) (33.85)351.74 -0.05 0.949 Pre vs Post DG PPC left 23.21 (5.98) (6.05)23.19 -0.02 (3.23)23.16 23.14 (2.98) -0.02 0.975 Pre vs Post DG PPC right 23.40 (6.33) (6.13)22.74 -0.66 (3.22)23.33 22.65 (2.78) -0.68 0.948 Pre vs Post DG Proportion cor-rected left 0.0004 (0.0005) 0.0004 (0.0005) 0.000002 0.0003 (0.0004) 0.0003 (0.0004) -0.000003 0.426 Pre vs post DG Proportion cor-rected right 0.0004 (0.0006) 0.0004 (0.0005) -0.000002 0.0003 (0.0004) 0.0003 (0.0004) -0.000005 0.565
Table 4 In order to further investigate the negative volumetric trend of the
hippocam-pus and the DG of both hemispheres a split plot ANOVA was conducted for the high vs low intensity exercise groups, comparing pre vs post exercise. We did this in order to reveal whether the direction of the volumetric change depends upon exercise intensity and which method is used to correct for ICV. Therefore, the a) raw volumes, b) the standard method, (Proportion method, VOI/ICV) and c) the Power Proportion Correc-tion method (PPC, VOIPPC=VOI/ICV𝛽 , (Liu et al., 2014)) were compared. Irrespective
of the method, all volumetric changes of the high and low intensity exercise groups are non-significant and close to zero. Thus, in both exercise intensity groups there were little to no change in volumes of the hippocampi and the DG over time. Neither the raw vo-lumes, the standard method (Proportion method) nor the PPC corrected volumes showed significant differences for the high and the low intensity groups over time. However, whether the change in volume is negative or positive seems to depend on which method is used to correct for ICV. For example, when the raw data and the PPC were used to calculate volumetric differences for the hippocampi of the high intensity group, a negative change in volume becomes apparent in comparison to the (small) posi-tive trend that emerges from the calculation with the Proportion method. This compari-son shows that the direction of volumetric changes could depend on which method is used to correct for ICV and should therefore be taken into consideration for further analysis.
Freesurfer vs. ASHS left hippocampus (mm³) Dif fe re nc e -3.000 -2.250 -1.500 -750 0 750 1.500 2.250 3.000 Mean 2.600 3.000 3.400 3.800 4.200 -2SD +2SD
Freesurfer vs. ASHS right hippocampus (mm³)
Dif fe re nc e -2.250 -1.500 -750 0 750 1.500 2.250 3.000 Mean 2.500 2.925 3.350 3.775 4.200 -2SD +2SD
Fig.5. Bland-Altman mean difference plots for hippocampal volumes between
Freesurfer (v.6) and ASHS. A) Freesurfer vs. ASHS left hippocampus. B) Freesurfer vs. ASHS right hippocampus. The horizontal axis shows the average of the two automated segmentation methods, and the their difference is depicted on the vertical axis. Volumes are in cubic millimeters.
Fig.6. Examples of the human hippocampus subfield segmentation of one subject;
Freesurfer v.6.0 (A) versus ASHS (B). The left hemisphere is depicted on the left. The top panel shows the coronal view, followed by the sagittal and axial view.
