Academic Editor: Robert Friedman
Background: Cognitive interventions (CIs) in the elderly are activities
that seek to improve cognitive performance and delay its deterioration. Our
objectives were to study potential genetic predictors of how a CI program may
influence immediate and delayed episodic verbal memory (EVM). Methods:
162 participants were elderly individuals without dementia who were randomized
into parallel control and experimental groups. Participants underwent genetic
testing to analyze the PICALM, ACT, NRG1, BDNF and APOE genes. We performed a broad
neuropsychological assessment before and 6 months after the CI. The CI involved
multifactorial training (30 sessions). The control group undertook the centre’s
standard activities. The main outcome measures were the genotype studied as a
predictor of post-intervention changes in EVM. Results: We found the CI
was associated with improvements in several cognitive functions, including
immediate and delayed EVM. While no individual gene was associated with any such
change, the interaction between PICALM/ACT (p = 0.008;
Eta
Age-related cognitive changes have been paid special attention since it has been recognized that Alzheimer’s disease (AD) commences many years before the first symptoms of minor memory loss become evident [1]. Specific activities have been developed that aim to improve memory and cognitive performance, increasing brain reserves while dampening cognitive decline and the evolution of dementia. These activities are generally referred to as Cognitive Interventions (CIs) and they have been the subject of intense development over the past two decades. These interventions usually have positive results and the effect sizes are commonly medium to medium-high when they fulfill certain conditions [2].
One of the important questions regarding CIs is whether elements exist that can predict their performance. What are the characteristics of people who benefit from training [3]? Few studies have been conducted on the modulation by genetic factors in the field of CI. However, of those that have, most focused on APOE, BDNF and genes involved with dopamine metabolism, catabolism and uptake.
It has been hypothesized that genes associated with cognitive improvements
enhance neural plasticity in certain brain regions, facilitating stronger
positive change. Polito et al. [4] implemented a cognitive
stimulation program observing improvement in memory and finding that improvement
only occurred in normal subjects who were non-carriers of APOE
BDNF (Brain Derived Neurotrophic Factor) is another gene that has been studied in relation to CIs. This is a growth factor that influences glutamatergic and GABAergic neurons, and its mechanisms of action and differential expression have been studied in regions thought to be quite important for memory, such as the hippocampus and prefrontal cortex [8]. In a study of a group of healthy senior citizens to determine whether improvement in attention and cognitive flexibility through training was conditioned by BDNF [9], it was found that homozygous Val/Val carriers of the most intensely studied single nucleotide polymorphism (SNP) of this gene, rs6265, achieved better results than Met/carriers. Other positive relationships with BDNF have been reported [10], and when the effects of APOE and BDNF on cognitive and physical training were studied, only APOE predicted improvements in verbal fluency perhaps because this was the most demanding task tested [11]. In this case an APOE and BDNF interaction was not considered and in fact, very few studies have contemplated gene associations and interactions in this field, reflecting the need for more efforts along those lines [12].
To date, we have been unable to find any studies on the possible use of
PICALM, NRG1 and ACT to predict the benefits of
training. However, their effects on neurotransmitters and beta-amyloid
(A
A few authors have proposed that the effects of some genes may be more readily detected in situations of demand, change or cognitive effort, such as that occurring in programs of cognitive training or stimulation [20]. However, since the effect sizes of each individual SNP are often low, it may be that the best way to detect any effect caused by them is through their interactions with other genes or SNPs. As a result, we contemplated this possibility in our study. This article is part of a randomized control trial in which multifactorial cognitive training was offered to healthy individuals. We found that, the training program followed, improved several cognitive areas including immediate and delayed episodic verbal memory (EVM). Thus, we set out to examine whether the genes PICALM (SNP rs3851179), ACT (SNP rs4934), BDNF (rs6265; Val66Met), NRG1 (rs6994992), and APOE were associated with the improvements observed in immediate and delayed EVM after training in the experimental group. We hypothesized that while we might not detect single gene effects, we may detect gene interaction effects. Although the various genes and their DNA variants may be conditioned by age, sex, cognitive reserve and other variables, our sole purpose was to study the effect of these genes regardless of these other variables.
This trial was retrospectively registered on January 29, 2020 (ClinicalTrials.gov -NCT04245579).
This study was performed on 226 older people,
The study was approved by the Clinical Investigation Ethics Committee at the HCSC (internal code No. 15/382-E_BS) and it was registered retrospectively on January 29th, 2020 (ClinicalTrials.gov-NCT04245579). Participants were informed of the study’s characteristics and signed consent forms. Every participant completed a medical record form and was given a structured interview, experienced professionals (a psychiatrist, neurologist and neuropsychologists) evaluating them in three, one-hour sessions. Participants were also examined by Magnetic Resonance Imaging (MRI) and Magnetoencephalography, although these data were not used here. The participants were assigned randomly to either of the two groups, Experimental and Control group, by an independent neuropsychologist blind to the individual process. The participants in the experimental group completed a cognitive intervention program based on the UMAM method, whereas no specific cognitive training was offered to the members of the control group. All the study subjects were assessed for a second time 6 months after the beginning of the study. Both groups followed all the routine activities that those who attend the center are offered (planned consultations, conferences-dialogue, general health recommendations…).
Each participant was assessed with: the Mini Mental State Examination (MMSE) [21]; the 7Minutes Test (7MT) [23]; Tests of Verbal Memory (Wechsler Logical Memory-EVM and DVM- and the Word List of the Wechsler Memory Scale-III -WMS-III) [24]; the Trail Making Test forms A and B (TMT) [25]; the Rule Shift Cards 1 and 2 from the Behavioral Assessment of the Executive Syndrome (BADS) [26]; the Stroop Test [27]; Rey’s Simple Figure Test (form B) [28]; the Rivermead Behavioral Memory Test (RBMT) [29]; the Boston Naming Test (BNT) [30]; the Semantic Verbal Fluency Test and the Phonological Fluency Test [31, 32]; and the Functional Activities Questionnaire (FAQ) [22]. Cognitive reserves were assessed using a specific questionnaire (CRQ) [33].
Genes and SNPs analyzed were PICALM (Chromosome 11; rs3851179, upstream SNP of the gen PICALM); ACT, also known as SERPINA3, (Chromosome 14; rs4934, G (ALA) changes to A (THR) in position 6 of protein SERPINA3 (Mutation missense); BDNF (Chromosome 11; rs6265, G (VAL) changes to A (MET) in position 66 of protein BDNF (mutation missense); NRG1 (Chromosome 8; rs6994992 in the 5-prime promoter region of the NRG1 gene regulates expression of the NRG1 type IV isoform); APOE (Chromosome 19; rs429358, T (CYS) changes to C (ARG) in position 130 of Apolipoprotein E (mutation missense), and rs7412, C (ARG) changes to T (CYS) in position 176 of Apolipoprotein E (mutation missense).
Genomic DNA was extracted from 10 mL EDTA-anticoagulated whole blood samples of
healthy elders. The detection of APOE genotype was
performed using a method revised from Zivelin et al. [34], although
utilizing the forward primer from the technique defined by Hixson and Vernier
[35] that produces a 227 bp DNA fragment. Double digestion of this fragment with
AflIII and HaeII yields on 4% agarose electrophoresis three specific fragments:
177 bp for
The CI program applied to the experimental group is a cognitive program initially designed and implemented in 1994 by the Memory Training Unit of the Madrid City Council. This program is known as the UMAM method and it has been described in a Manual of Evaluation and Memory Training published with explanations of each session and exercise, including booklets, printed sheets, other edited books, slides, etc. [36]. Briefly, the cognitive training program consists of 30 sessions, each 90 minutes long, and the sessions are carried out in groups of 12–18 people. This cognitive program is currently being implemented in 16 municipal health centers run by the Madrid City Council and it has been applied to more than 30,000 individuals.
All the statistical analyses were performed using the IBM SPSS Statistics
package (IBM Corp. Released 2011. IBM SPSS Statistics for Windows, Version 20.0.
Armonk, NY). To analyze the treatment effect in both immediate and delayed EVM, a
General Linear Model with repeated measures was used (post-training minus
pre-training). The within-subject factor between the pre- and post-training
evaluation, the two-time points examined, was designated as “Time,” and the
between-subject factor, that is the experimental or control group, was designated
as “Group”. The difference in response between the experimental and control
group was reported as the “Time
To study the predictors in the experimental group, we first used an ANOVA test
with a partial Eta squared (Eta
For those variables with a significant interaction, we analyzed the effect of the alleles and every single gene interaction on the dependent variable using the “Decision Tree” procedure. The gene with the highest F value found by ANOVA was used as the first variable to split the sample in the analysis. It has been pointed out that this procedure allows us to identify different risk profiles and that such partitioning approaches do not imply interaction “per se” but potential interactions [38], which was the meaning we wanted to give to the results of our decision tree.
We used raw scores of all the data in the statistical analysis and the
statistical significance was set at p
In this study 226 people aged over 60 years were assessed for eligibility of
which 211 were randomized to the two study groups. 38 participants were excluded
because of invalid blood samples and 11 because of poor motivation to continue in
the study. The final study sample was comprised of 162 individuals: mean
age = 71.04
Experimental group | Control group | Time | Group | Interaction (group | |||||||
n = 88 | n = 74 | ||||||||||
Variable (range) | Pre-training | Post-training | Pre-training | Post-training | F | p | F | p | F | p | Cohen’s d |
M (SD) | M (SD) | M (SD) | M (SD) | ||||||||
WLM Immediate (0–75) | 34.38 (12.24) | 38.59 (11.97) | 35.92 (12.89) | 36.55 (12.76) | 16.28 | 0.000 | 0.02 | 0.903 | 8.90 | 0.003 | 0.26 |
WLM Delayed (0–50) | 19.14 (9.73) | 23.30 (9.60) | 21.8 (10.38) | 22.75 (9.75) | 27.29 | 0.000 | 0.49 | 0.487 | 11.60 | 0.001 | 0.34 |
WLM, Wechsler Logical Memory [Immediate Episodic Verbal Memory (EVM) and Delayed Episodic Verbal Memory (EVM)]. |
The genetic data was obtained from all the participants (for the experimental group: Table 2) and the genotype frequency in the cohort did not deviate from the Hardy–Weinberg equilibrium: NRG1 (p = 1); BDNF (p = 0.47); PICALM (p = 0.49); ACT (p = 0.07); APOE rs429358 (p = 1); APOE rs7412 (p = 1).
Gene | N | % | Gene | N | % | |
NRG1 | BDNF | |||||
CC | 23 | 26.4 | AA | 4 | 4.6 | |
CT | 43 | 49.4 | AG | 24 | 27.6 | |
TT | 21 | 24.1 | GG | 59 | 67.8 | |
ACT | PICALM | |||||
GG | 27 | 31.4 | AA | 10 | 11.4 | |
AG | 50 | 58.1 | AG | 45 | 51.1 | |
AA | 9 | 10.5 | GG | 33 | 37.5 | |
APOE (haplotype) | ||||||
e2e3 | 7 | 8.1 | ||||
e3e3 | 60 | 69.8 | ||||
e3e4 | 18 | 20.9 | ||||
e4e4 | 1 | 1.2 |
We analyzed the effect of each SNP on the basal immediate and delayed EVM scores
of experimental group, and no significant associations were observed, although
the effect sizes remained small. While none of the two-by-two gene interactions
produced any significant association either, the largest effect sizes (Eta
Association of Genes with basal data (Evaluation Pre) | ||||||
WLM Immediate | WLM Delayed | |||||
Single Genes | F | Sig. (p) | Eta |
F | Sig. (p) | Eta |
ACT | 0.297 | 0.744 | 0.007 | 0.939 | 0.395 | 0.022 |
BDNF | 2.221 | 0.115 | 0.050 | 2.259 | 0.122 | 0.049 |
NRG1 | 0.332 | 0.718 | 0.008 | 0.130 | 0.878 | 0.003 |
PICALM | 0.718 | 0.491 | 0.017 | 0.950 | 0.391 | 0.022 |
APOE | 0.086 | 0.770 | 0.001 | 0.031 | 0.860 | 0.000 |
Interactions | ||||||
NRG1 + BDNF | 1,017 | 0.426 | 0.084 | 1.230 | 0.297 | 0.099 |
NRG1 + ACT | 0.629 | 0.751 | 0.062 | 0.982 | 0.457 | 0.094 |
NRG1 + PICALM | 1.490 | 0.175 | 0.133 | 1.365 | 0.225 | 0.123 |
NRG1 + APOE | 0.166 | 0.974 | 0.010 | 0.255 | 0.936 | 0.016 |
BDNF + ACT | 0.772 | 0.613 | 0.066 | 0.848 | 0.551 | 0.072 |
BDNF + PICALM | 1.537 | 0.167 | 0.120 | 1.652 | 0.133 | 0.128 |
BDNF + APOE | 1.154 | 0.337 | 0.055 | 1.178 | 0.327 | 0.056 |
PICALM + APOE | 1.291 | 0.276 | 0.075 | 0.867 | 0.507 | 0.051 |
ACT + APOE | 0.318 | 0.901 | 0.020 | 0.502 | 0.774 | 0.031 |
ACT + PICALM | 0.460 | 0.860 | 0.040 | 0.652 | 0.712 | 0.055 |
WLM, Wechsler Logical Memory. |
We studied each of the five genes to analyze their association with the
post-training changes in immediate and delayed EVM (Table 4). No significant
association was observed for any single gene and the effect sizes were very
small. The largest effect sizes for immediate EVM were evident with the
NRG1 + PICALM interaction, as was the case for the NRG1
+ ACT interaction, which was also associated with changes in
delayed EVM. Upon studying PICALM in more detail, and
focusing on the AA allele as opposed to the AG/GG allele, a significant
association with EVM was found (F = 4.432; p = 0.038;
Eta
Association of Genes with Change by training (Differences Pre-Post) | ||||||
Change in WLM | Change in WLM | |||||
Immediate Pre-Post | Delayed Pre-Post | |||||
Gene | F | Sig. (p) | Eta |
F | Sig. (p) | Eta |
NRG1 | 2.605 | 0.080 | 0.064 | 2.451 | 0.093 | 0.061 |
BDNF | 0.037 | 0.964 | 0.001 | 0.089 | 0.915 | 0.002 |
ACT | 1.457 | 0.239 | 0.037 | 2.462 | 0.092 | 0.062 |
PICALM | 2.188 | 0.119 | 0.054 | 0.541 | 0.585 | 0.014 |
APOE | 0.000 | 0.994 | 0.000 | 0.081 | 0.776 | 0.001 |
Interactions | ||||||
PICALM + ACT | 3.002 | 0.008 | 0.231 | 1.461 | 0.195 | 0.127 |
PICALM + APOE | 1.412 | 0.230 | 0.088 | 0.206 | 0.959 | 0.014 |
NRG1 + BDNF | 0.953 | 0.463 | 0.075 | 3.114 | 0.009 | 0.208 |
NRG1 + ACT | 1.656 | 0.125 | 0.163 | 1.326 | 0.246 | 0.135 |
NRG1 + PICALM | 2.398 | 0.029 | 0.191 | 1.297 | 0.264 | 0.113 |
NRG1 + APOE | 1.206 | 0.315 | 0.077 | 0.918 | 0.474 | 0.060 |
BDNF + ACT | 0.650 | 0.713 | 0.062 | 1.137 | 0.351 | 0.103 |
BDNF + PICALM | 0.959 | 0.468 | 0.086 | 0.609 | 0.747 | 0.057 |
BDNF + APOE | 0.243 | 0.421 | 0.013 | 0.138 | 0.968 | 0.008 |
ACT + APOE | 0.968 | 0.443 | 0.064 | 1.186 | 0.325 | 0.077 |
WLM, Wechsler Logical Memory. Bold data correspond to significant associations. |
PICALM + ACT (WLM Change Immediate) | |||||
PICALM | (I) ACT | (J) ACT | Mean Difference (I-J) | Std. Error | Sig. (p) |
GG | AA | AG | 14.471* | 5.067 | 0.017 |
GG | 17.364* | 5.210 | 0.004 | ||
AG | AA | –14.471* | 5.067 | 0.017 | |
GG | 2.893 | 2.623 | 0.821 | ||
GG | AA | –17.364* | 5.210 | 0.004 | |
AG | –2.893 | 2.623 | 0.821 | ||
NRG1 + PICALM (WLM Change Immediate) | |||||
NRG1 | (I) PICALM | (J) PICALM | Mean Difference (I-J) | Std. Error | Sig. (p) |
CT | AA | AG | –11.202* | 3.794 | 0.013 |
GG | –11.375* | 3.888 | 0.014 | ||
AG | AA | 11.202* | 3.794 | 0.013 | |
GG | –0.173 | 2.308 | 1.000 | ||
GG | AA | 11.375* | 3.888 | 0.014 | |
AG | 0.173 | 2.308 | 1.000 | ||
NRG1 + BDNF (WLM Change Delayed) | |||||
NRG1 | (I) BDNF | (J)BDNF | Mean Difference (I-J) | Std. Error | Sig. (p) |
CT | AA | AG | –11.202* | 3.794 | 0.013 |
GG | –11.375* | 3.888 | 0.014 | ||
AG | AA | 11.202* | 3.794 | 0.013 | |
GG | –0.173 | 2.308 | 1.000 | ||
GG | AA | 11.375* | 3.888 | 0.014 | |
AG | 0.173 | 2.308 | 1.000 | ||
* The mean difference is significant at the 0.05 level.
Adjustment for multiple comparisons: Bonferroni. |
We then analyzed the gene interactions that might predict a change with training
in immediate EVM (pre-post). The association between the genes PICALM +
ACT was significant, both in terms of the effect of each gene individually as
well as their interaction, with an overall effect size of 0.23 (see Table 4).
Upon introducing APOE into the PICALM + ACT model and studying
the interactions between the three genes, the equation continued to be
significant (F = 2.062; p = 0.029; Eta
To determine whether any of the alleles might be associated with changes in immediate and delayed EVM, we followed the “Decision Tree” procedure using the groups of variables for which a significant interaction was identified. Analyzing PICALM and NRG1 (Fig. 1), we can observe that when the mean improvement in “immediate EVM” was assessed for the whole group (Fig. 1), the performance of individuals with the AA alleles (node 1) did not improve but rather worsened (note that the negative sign mean indicates a decline in performance, –0.875: Fig. 1). The interaction of the PICALM AG and GG carriers with the participants carrying the NRG1 CT alleles performed three times better than those with the CC and TT alleles, who also performed worse than the mean of the entire sample (node 4) (Fig. 1).
Decision Tree Change in WLM immediate as dependent variable split by alleles of the two genes with interaction significant effects. Over each node the correspondent PICALM and NRG1 alleles is shown, inside the node is the mean of change in the dependent variable, with its SD, the total “n” in the node, and the percentage (%) of participants in this node with respect to the sample of each specific gene; p-value; df, degrees of freedom.
Interacting with PICALM (Fig. 2), the carriers of the AA and GA, ACT variants (node 3) improved nearly three times more than carriers of the GG alleles (node 4).
Decision Tree Change in WLM immediate as dependent variable split by alleles of the two genes with interaction significant effects. Over each node the correspondent PICALM and ACT alleles is shown, inside the node is the mean of change in the dependent variable, with its SD, the total “n” in the node, and the percentage (%) of participants in this node with respect to the sample of each specific gene; p-value; df, degrees of freedom.
On studying the change in the delayed EVM (Fig. 3), we observed that the CT carriers of the NRG1 gene (node 2) performed better. Regarding the BDNF gene and its interaction with the NRG1 gene, carriers of the AG alleles of BDNF showed the greatest improvement after training in node 6, whereas in node 4 the carriers of this BDNF allele (AG) showed a decline in performance. Carriers of BDNF AA were not studied here as they comprised only 4.6% of the sample and there were no carriers of NRG1 CT who were also BDNF AA.
Decision Tree Change in WLM immediate as dependent variable split by alleles of the two genes with interaction significant effects. Over each node the correspondent NRG1 and BDNF alleles are shown, inside the node is the mean of change in the dependent variable, with its SD, the total “n” in the node, and the percentage (%) of participants in this node with respect to the sample of each specific gene; p-value; df, degrees of freedom.
We studied here a group of elderly individuals without dementia who were randomly assigned to either a control or experimental group. The experimental group participated in a memory training and cognitive stimulation program based on the UMAM method, which achieved improvements in immediate and delayed EVM, as measured by the WLM (immediate and delayed scoring). We set out to analyze whether the ACT, PICALM, BDNF, NRG1 and APOE genes could serve as predictors of this change or the improvement following training. The results showed that none of these genes were associated with the participants’ basal scoring in immediate or delayed EVM, neither individually or via two-by-two interactions. Furthermore, there was no association between the individual genes and the change induced by training. However, when the interaction between genes was studied, the interaction between PICALM and ACT was significantly associated with the change in immediate EVM (beneficial alleles: PICALM AG/GG and ACT AA/AG), as was the interaction between PICALM and NRG1 (beneficial alleles: NRG1 CT, PICALM AG/GG). The interaction between NRG1 and BDNF proved to be significantly associated with the change in delayed EVM. The APOE gene did not appear to be significantly associated with any improvement or decline in EVM performance, either alone or in any interaction, although the data available may not have had sufficient statistical power to fully analyze all such interactions. It is interesting to note that the NRG1 gene is involved with two interactions (with PICALM and BDNF) associated with an improvement in immediate and delayed EVM, and that the same CT allele is involved in the interactions with both genes, indicating it could be considered a “beneficial” allele in terms of such tasks. Interestingly, we have found that some heterozygotes perform better than homozygotes, which is less usual [39, 40].
In terms of the basal scores of the participants for immediate or delayed EVM, associations between cognitive (“basal”) performance and some genes have been found previously in adults [41], and the inheritability of general memory is usually considered to be around 50% [42]. However, elsewhere studies of post-training predictors have found similar results to those seen here, as no association was found between the basal scores and the genes studied. When training working memory in children, two SNPs were seen to predict the results but none were associated with basal performance [12]. No such association was found for a single gene in an earlier study on EVM, although combinations of genes did seem to have some effect [39]. While this failure to detect an association may be due to the lack of any association, it may also just reflect the low effect size [43]. Nevertheless, the best results are generally found in gene association studies.
We did not find an APOE allelic variant to be a predictor of
improvement in EVM despite the supposed evidence that non-carriers of the
In a previous study of the PICALM, CR1, BIN1,
CLU and APOE genes, Barral et al. [39] did not find
any single SNP associated with EVM in a non-AD population, although some
association with poorer performance was detected for some alleles in interactions
with other genes: e.g., PICALM GG, CR1 GG and APOE
In relation to the ACT gene and the rs4934SNP, it is unclear whether there is an association with risk or protection. It has been proposed that the onset of AD is earlier in subjects with the GG alleles and that their life expectancy is lower [48], yet in a different study AA was considered to be a risk allele [49]. The most severe association was also reported to be the ACT 5 G allele plus the ACT 7 C allele [50]. Here, the ACT allele associated with worst performance was the GG allele (which would be considered the risk allele), conditioned by PICALM AG/GG.
How can the PICALM/ACT interaction be explained? Again, a
large number of studies of these genes are related to AD in which the events
underlying this interaction remain unclear. Both genes are expressed in regions
of the brain linked to AD, both are associated with actions on A
We did not find any studies into interactions between PICALM and NRG1 in relation to cognitive performance in any area. If we try and consider the nature of their interaction, important features of these two genes can be taken into account, such as their strong expression in the hippocampus and frontal cortex [52, 53]. Some of the actions of PICALM occur in the pre- and post-synaptic regions, regulating aspects of exocytosis/endocytosis. Indeed, PICALM has been implicated in the mechanisms by which neurotransmitter vesicles associate with the presynapse [54], a key aspect of the Long-Term Potentiation (LTP) linked to memory regulation by NRG1 [17].
The NRG1 CT alleles could be a “beneficial” variant in delayed EVM when this gene is associated with BDNF, whereas the improvement associated to the CC and TT alleles represents about half that of the heterozygote, representing risk alleles. We did not find any other study of this gene interaction effect, although many studies (some related to schizophrenia) have associated NRG1 with structural differences in the brain, identifying the TT alleles as risk alleles associated with a loss of gray and white matter in several regions of the human brain [55].
Altered BDNF activity in the hippocampus can lead to a decline in
memory tasks [56] and in most cases, the adenine allele (encoding Met) is
associated with worse cognitive performance than the guanine allele (encoding
Val). This may be associated with the reduction in hippocampal volume, as occurs
in subjects with a high level of A
As far as the effects following training are concerned, BDNF was not seen to be a predictor in a previous study [12], although a predictor effect was observed in our results when interacting with NRG1. In an attempt to understand this finding, we can consider the relationships between these two genes. Some of their activities are similar or complementary, especially their effects on neural plasticity and LTP [17]. The NRG1 receptor ErbB4 activates a BDNF protein receptor, TrkB, indicating that an increase in NRG1 will also increase the levels of BDNF. Both NRG1 and BDNF induce myelination in the central nervous system (CNS) by increasing the number of NMDARs, and BDNF plays a compensatory role in this myelination when the activity of NRG1 is dampened or absent [57]. The hippocampus and frontal region, common predominant sites of their activity, are important areas for memory and particularly, delayed memory. BDNF has been found to enhance cognitive performance post-physical or cognitive exercise [58] and NRG1 levels may increase with neuronal activity [59]. Despite these possible functional and physical interactions, we have not found a truly plausible explanation for the specific interaction observed in the improvement of delayed memory. Might this interaction influence LTP? If the NRG1 CC/TT alleles were to reduce the levels of BDNF in AG carriers, it may have an adverse effect on the post-training influence on delayed EVM memory. Conversely, the effects may be enhanced in the two heterozygous types as the NRG1 CT could produce an increase of BDNF in subjects who are carriers of AG alleles.
There are several limitations to our study that should be highlighted. We are aware that some factors that are associated with an improvement or a decline were not analyzed in this work (cognitive reserve, brain volume, education level, motivation, etc.). While this may be viewed as a limitation, we wanted to focus our work about CI specifically on the influence of genotype, irrespective of other factors, and even their possible influence on the expression of these genes. Given the number of participants, it is difficult to conduct studies on the interactions of more than two genes and moreover, this was a possible cause of the inconsistent results from BDNF AA carriers. Furthermore, our explanations for the physiological effect of the genes must be understood as possibilities or hypotheses that require further study. As such future efforts should be specifically designed to confirm or reject the relationships identified here.
Recent genetic studies, including genotyping in specific diseases [60], have highlighted the role of oxidative stress as an inducer of neurodegeneration, causing toxicity and promoting cell death [61, 62]. Learning and memory are possible only because of the neural plasticity; genetic and environmental factors are involved both in pathological processes and in slowing or preventing cognitive decline; molecular biomarkers can contribute to the diagnosis of neurodegeneration and guide pharmacological interventions [63]. Here we aimed to open new avenues in the studies of the influence of specific genes on the benefits of non-pharmacological therapies.
Our study reinforces the concept that gene interactions may be better studied in situations of stress or high demand as their expression could be boosted in such situations. However, this study has practical applications as while a large proportion of the sample improved with the CI in terms of the variables analyzed, immediate and delayed EVM, some participants did not change or even achieved lower scores when assessed six months after training. Therefore, we must adapt the cognitive training and stimulation to each participant’s characteristics, including their genotype, which has not been intensely studied. If the results obtained are taken into consideration, it should be possible to achieve better results with these Health Prevention and Promotion activities.
PMC, MMP and DPC, were responsible for the design of the study, carried out the statistical data analysis and wrote the paper; CMR and MLDL supervised the data collection, revised the manuscript for important clinical and scientific content. All authors read and approved the final manuscript.
The study was approved by the Clinical Investigation Ethics Committee of the Clinic Hospital San Carlos (HCSC) (internal code nº 15/382-E_BS). Participants were informed of the study characteristics and signed consent forms. Trial registration: Trial retrospectively registered on January 29, 2020; ClinicalTrials.gov-NCT04245579.
Not applicable.
This work was supported by the Ministry of Science and Innovation of Spain: PSI 2015-68793-C3-3-R and PSI 2015-68793-C1-3-R.
The authors declare no conflict of interest.
Publisher’s Note: IMR Press stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.