Long-Term Changes in Brain Derived
Neurotrophic Factor and Cognition in Late-Life
Depression Patients Treated with
Electroconvulsive Therapy
Research Project & Thesis | February 2020 – June 2020 |Bachelor Psychobiology
Written by: E.M. van Vliet
Student ID: 11682418
Supervisors: M.J. Wagenmakers and M.L. Oudega
19-6-2020
1 Abstract
Background. Late-life depression (LLD) has been associated with cognitive impairment and a
decrease in brain derived neurotrophic factor (BDNF). In addition, electroconvulsive therapy (ECT) has been shown to improve cognition and increase BDNF levels. Moreover, BDNF plays an important role in neuronal plasticity and consequently has an effect on cognition. Possibly, BNDF can explain the changes in cognition after ECT. Therefore, this study measured serum BDNF levels until 4 weeks after ECT and global cognitive functioning with the Mini Mental State Examination (MMSE) in severe LLD patients with a follow-up period of 5 years after ECT.
Methods. Based on the baseline BDNF levels a low and high BDNF group were created. Subsequently,
longitudinal changes in BDNF and MMSE scores in these groups were examined. Lastly, the correlation between BDNF group and longitudinal MMSE scores in severe LLD patients was tested.
Results. BDNF levels increased significantly in the low BDNF group (F(1, 157) = 10.5, p = .001). MMSE
scores increased significantly after ECT in both the low (F(1, 223) = 16.1, p < .001) and high (F(1, 202) = 7.27, p = .008) BDNF group. No association was found between BDNF group and the changes in MMSE scores during and after ECT (F(5, 357) = 0.51, p = .77).
Conclusion. BDNF levels increased after ECT in patients with low BDNF levels at baseline. Global
cognitive functioning improved up to 6 months after ECT, but this improvement was not correlated with BDNF group. Thus, the cause of cognitive changes in severe LLD patients treated with ECT remains unclear.
2 Introduction
Depression in people over 55 years old (late-life depression or LLD) has a prevalence of 16.5% (Volkert, Schulz, Härter, Wlodarczyk & Andreas, 2013) and is characterized by treatment resistance, cognitive impairment and somatic comorbidities (Naismith, Norrie, Mowszowski & Hickie, 2012). People with LLD show a decrease of functioning in the cognitive domains of episodic memory, processing speed and interference control (Korten et al., 2014). As a result, around 50% of depressed elderly meet criteria for mild cognitive impairment (MCI) (Adler, Chwalek & Jajcevic, 2004; Lee, Potter, Wagner, Welsh-Bohmer & Steffens, 2007). In addition, brain derived neurotrophic factor (BDNF) levels are lower in people with a depression compared to healthy controls. (Molendijk et al., 2014). Since BDNF stimulates neuronal plasticity and cell growth, plasticity is expected to be lower in depressed patients. According to the neurotrophic hypothesis the reduction in BDNF is caused by stress and the decrease is mainly found in brain regions that regulate emotion and memory. The subsequent decrease in neuronal plasticity in those regions has been linked to depression (Bus & Molendijk, 2016). However, some have suggested lower BDNF levels are a result of depression, not a cause. For example, Bus et al. (2015) found BDNF decreased after a long period of depression, while patients with a short depression had normal levels of BDNF. Furthermore, antidepressants are related to an increase of BDNF, a normalization of neuronal plasticity and they have a positive influence on depression. Nevertheless, it is not known what the causal relation is between
depression and BDNF (Bus & Molendijk, 2016). In addition, the cause of cognitive impairment in LLD is unknown. Possibly, BDNF can explain the cognitive changes in LLD, since BDNF and neuronal plasticity are closely related.
Although BDNF and cognitive functioning are decreased in LLD, they can improve again after treatment. In LLD the preferred treatment is electroconvulsive therapy (ECT) because of higher response rates (60-70%) (Spaans et al., 2015) compared to antidepressants, which have an effectiveness of around 50% after the first treatment (Naismith et al., 2012). ECT involves a brief electrical stimulation of the brain. Although the exact working mechanism of ECT is not yet known, it has been found to increase BDNF. Bocchio-Chiavetto et al. (2006) found a significant increase of BDNF one month after ECT and having lower BDNF levels before treatment is associated with higher chance of remission after ECT (van Zutphen et al., 2019). ECT can also improve cognition. An improvement of the Mini Mental State Examination (MMSE), a test for global cognitive functioning, has been found during ECT, which remained stable until 6 months after (Obbels et al., 2019). This improvement in cognition after ECT is seen both in people without cognitive impairment and in people with MCI before ECT (Hausner, Damian, Sartorius & Frölich, 2011).
3 Thus, both BDNF and cognition are affected in depression and because of the role BDNF plays in neuronal plasticity, BDNF might be the underlying mechanism affecting cognition. This relationship has been confirmed in healthy controls, where higher BDNF was correlated with a higher MMSE score (Gunstad et al., 2008). The association has also been found in Alzheimer patients, where higher BDNF levels were correlated with slower cognitive decline (Laske et al., 2011). Another study found cognition to be associated with BDNF during treatment with antidepressants in depressed adults. They found that normalization of memory problems was related to higher BDNF plasma levels (Engelmann et al., 2019). However, the study of Dols et al. (2015) did not show an association between BDNF and cognition in depressed elderly. They suggest BDNF levels might not decrease in stages of the disease where remission is still possible, since BDNF is crucial for neuronal cell survival and functioning. Perhaps, the association is only present in more severe cases of depression. Since people who are treated with ECT are more severely depressed, it is possible that in these patients, a relation between BDNF and cognition is present.
Therefore, in this study the longitudinal changes of BDNF levels and global cognitive functioning induced by ECT were investigated. It was also tested whether there was an association between BDNF levels at baseline and the changes in global cognitive functioning during and after ECT in patients with severe LLD. The goal was to gain more insight in the underlying mechanism of the cognitive changes in severe LLD patients treated with ECT.
Based on previous studiesit was expected that both BDNF levels (Bocchio-Chiavetto et al., 2006) and cognitive functioning (Obbels et al., 2019) would increase after ECT. Since BDNF and cognition are positively associated in healthy controls (Gunstad et al., 2008), Alzheimer patients (Laske et al., 2011) and in depressed adults during treatment with antidepressants (Engelmann et al., 2019), it was hypothesized that baseline BNDF levels would be positively related to cognition during and after ECT.
4 Methods
Study sample
For this study data were obtained from the Mood Disorders in Elderly treated with Electroconvulsive Therapy dataset (MODECT, Dols et al., 2017), a study including 110 patients older than 55 years with severe unipolar depression, according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR, American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 2000), who were referred for ECT. Patients were recruited from two tertiary psychiatric hospitals (GGZ inGeest, Amsterdam, the Netherlands and University Psychiatric Center, KU Leuven, Belgium) between 1 January 2011 and 31 December 2013. Exclusion criteria were a DSM-IV-TR diagnoses of bipolar disorder, schizoaffective disorder or a history of major neurologic illness (including Parkinson disease, stroke, and dementia). A psychiatrist made the diagnoses and diagnoses were confirmed by the Mini International Neuropsychiatric Interview (Shéehan et al., 1998). The study protocol of MODECT was approved by the Ethical Review Board of the VU University Medical Center and the Ethical Review Board of the Leuven University Hospitals and conducted according to the Declaration of Helsinki (clinicaltrials.gov; NCT02667353). Written informed consent was obtained from all patients at the start of the baseline assessment. In the current study, 24 patients were excluded, because of missing BDNF data at baseline, rendering to a total of N = 86 in the current study.
ECT procedure
Patients received ECT twice a week in accordance with Dutch standards (Birkenhager et al., 2010). ECT was administered with the Thymatron System IV (Somatics, LLC, Lake Bluff, Il, USA) (maximum energy 200%, 1,008 C). At first treatment, stimulus intensity was determined by empirical dose titration. For right unilateral ECT it was six times the initial seizure threshold and for bilateral ECT 1.5 times the seizure threshold. All patients were treated with brief-pulse ECT (0.5-1.0 ms). If the duration of a motor seizure was less than 20 seconds, the dose was raised according to Dutch guidelines (Birkenhager et al., 2010). Patients were switched from unilateral to bilateral ECT when the clinical condition worsened (i.e., an increase in total Montgomery Asberg Depression Rating Scale (MADRS) scores, presence of debilitating psychotic features, increased suicidality, dehydration or weight loss, or when after six unilateral treatments there was no clinical improvement according to the judgment of the treating psychiatrist). ECT ended when a MADRS score of less than 10 was achieved at two consecutive ratings with a week interval or when no further improvement in clinical condition was shown during the last 2 weeks of ECT sessions after a minimum of six unilateral and six
5 bilateral sessions. If possible, psychotropic medication was discontinued at least 1 week before ECT. Otherwise, it was kept stable from 6 weeks before ECT and during the ECT course.
Clinical assessment
Demographic variables and medication use were obtained by interview. Depression severity was assessed weekly with the MADRS (Montgomery & Asberg, 1979). The MADRS is a 10-item
questionnaire, including questions about sadness, tension and concentration difficulties. The total score ranges from 0-60. A higher score indicates a more severe depression. Remission was defined as a MADRS score of less than 10 points at two consecutive ratings with a week interval and response as a decrease in MADRS scores of at least 50% after ECT compared to baseline.
Neuropsychological assessment
Cognitive functioning was assessed with the MMSE, a test for global cognitive functioning. The MMSE is a short test that measures orientation, attention, executive functioning, memory and language. The MMSE generates a score ranging from 0 to 30, where 30 is a perfect score. A score of 23 is often used as cut-off to identify people with dementia (Folstein, Folstein & McHugh 1975). The MMSE was conducted before ECT (baseline or T0), after the sixth ECT (T1), 1 week after the last ECT (T2), 4 weeks after the last ECT (T3), 6 months after ECT (T4) and 5 years after ECT (T5). The assessment of the MMSE and MADRS was carried out by trained research nurses who were blinded to clinical information.
BDNF measurements
Serum BDNF levels were measured through blood samples at T0, T1, T2 and T3. Blood samples were collected between 7.30 and 9.30 AM. after an overnight fast. Serum was separated and stored at -85 ⁰C. Before being assayed, it was stored for 7 to 43 months. The Emax Immuno Assay System from Promega (Madison, WI, USA), catalogue number G7610, was used to determine the BDNF protein levels in ng/ml in serum. To increase detectability, undiluted serum was acid-treated in a dilution-dependent manner. Before reading the absorbency in duplicate with the Bio-Rad (Hercules, CA, USA) Benchmark microplate reader at 450 nm, serum samples were diluted 100 times. None of the subjects had BDNF values below the reliable detection threshold of 1.56 ng/ml. Levels at baseline ranged from 6.04 to 35.59 ng/ml. A BDNF level of 23.3 ng/ml has been found to be average in healthy older adults (Ziegenhorn et al., 2007).
6
Statistical analysis
Demographics. RStudio version 1.2.5042 was used for all statistical analysis. First, a low BDNF group
and a high BDNF group were created. The median (17.9 ng/ml) was used as cut-off (van Zutphen et al., 2019). The sociodemographic characteristics of both groups were compared with χ2-tests for categorical variables and independent t-tests were used for normally distributed continuous variables. Yates’ continuity correction was applied when assumptions of the χ2-tests were not met. The Levene’s test showed the assumption of equal variances was met in all continuous variables. The assumption of normal distribution was tested with the Shapiro Wilk test, in the case of p-values lower than 0.05, data were transformed or, if that did not help, a Mann-Whitney U test was conducted.
Confounders and covariates. In the statistical tests assessing the changes in BDNF levels, variables
that were found to be significantly different between the low and high BDNF group were included as covariates. The additional covariates age and gender were chosen based on literature (Bus et al., 2012). To check for possible confounders in the current dataset, independent t-tests were performed on the independent variables that significantly differed between the low and high BDNF group to check whether they also affected MMSE scores. If so, they were included as a confounder in all further analysis. Tests looking at the changes in MMSE scores were corrected for the covariates age, gender, educational level and presence of psychotic symptoms, all based on literature (Obbels et al., 2019).
Baseline correlation. The correlation between BDNF and MMSE at baseline in the low and high BDNF
group was tested with Kendall’s tau-test, since BDNF levels were not normally distributed.
Longitudinal changes BDNF. To assess the changes in BDNF levels induced by ECT and the differences
in longitudinal BDNF data between the two BDNF groups, a repeated measures Analysis of Variance (ANOVA) was performed. With this test, the effects of time and BDNF group on BDNF levels were tested and the interaction effect of time and BDNF group was tested. Data were checked for
normality with the Shapiro Wilk Test, no abnormalities were found. P-values were corrected with the Greenhouse-Geisser correction in case of violation of sphericity. Post-hoc testing was performed with two repeated measures ANOVA’s. The first tested at which time points the BDNF levels differed between the two BDNF groups. The second determined in which BDNF groups the BDNF levels changed significantly over time. This led to an additional pairwise t-test for the low BDNF group to check which time points differed significantly. The Bonferroni correction for multiple testing was applied.
7
Longitudinal changes MMSE and correlation with BDNF. The changes in MMSE scores over time and
the differences in the longitudinal MMSE data between the low and high BDNF group were assessed with a repeated measures ANOVA. It was tested whether time and BDNF group had a significant effect on MMSE scores and whether there was an interaction effect of time and BDNF group. Small deviations from normality were still present in the longitudinal data after transformation, but because of its robustness, a repeated measures ANOVA was still possible. The Greenhouse-Geisser correction was applied in case of non-sphericity. For post-hoc testing, a repeated measures ANOVA was performed to test at which time points the two BDNF groups differed in MMSE scores. Another repeated measures ANOVA was conducted to test in which BDNF groups the MMSE scores changed over time. Based on those results pairwise-t-tests were performed for both BDNF groups to test which time points differed significantly. P-values were adjusted for multiple testing with the Bonferroni correction.
8 Results
Demographics. A total of 86 subjects was included, ages ranged from 55 to 92 (M = 72.9, SD = 8.4).
Most of the participants were female (Table 1). BDNF levels at baseline ranged from 6.04 to 35.59 ng/ml, with a median of 17.9 (IQR = 7.6). MMSE scores at baseline ranged from 7 to 30, with a mean value of 24 (SD = 5.17). For the baseline MMSE data a log transformation with the reversed data was needed for normal distribution. Consequently, for all further analysis MMSE data of all time points were reversed and log-transformed. Comparison of the main characteristics showed a few significant differences between the low and high BDNF group (Table 1). Compared to the low BDNF group, the high BDNF group included more subjects who used mood stabilizers during their current episode. In addition, there were more subjects included in Amsterdam in the high BDNF group, more subjects with a physical comorbidity and less remitted patients, which is in line with van Zutphen et al. (2019).
Confounders and covariates. With t-tests, no association was found between the variables that
differed between the BDNF groups and MMSE scores, so no confounders were found in the current dataset (data not shown). In the tests assessing BDNF data the covariates were the variables that differed between the BDNF groups, which were the following: inclusion site, use of mood stabilizers, presence of a physical comorbidity and remission status. In addition, the tests were corrected for age and gender. For tests looking at changes in MMSE scores the following covariates were chosen based on literature: age, gender, educational level and presence of psychotic symptoms.
Baseline correlation. BDNF and MMSE scores were not correlated at baseline in both the low (τ = .09, p = .41,) and high (τ = -.10, p = .37) BDNF group.
Longitudinal changes BDNF. In the longitudinal BDNF data (Figure 1), the repeated measures ANOVA
(Appendix, Table 1) showed there was a significant effect of BDNF group on BDNF levels and a significant interaction between BDNF group and time, meaning that the BDNF levels differed between the BDNF groups and that the BDNF groups differed in the way the BDNF levels changed over time after ECT. The repeated measures ANOVA also showed inclusion site and age were significantly correlated with BDNF levels. In the post-hoc analysis, the first repeated measures ANOVA showed the difference in BDNF levels between the two BDNF groups was present at all time points (Figure 1). The second repeated measures ANOVA showed BDNF levels changed significantly over time after ECT in the low BDNF group (F(1, 157) = 10.5, p = .001), but not in the high BDNF group (F(1,153) = 2.73, p = .10). With a pairwise t-test it was found that in the low BDNF group, the BDNF levels had increased significantly 4 weeks after the last ECT compared to baseline (p < .001) and compared to the BDNF levels after the sixth ECT (p = .01).
9
Longitudinal changes MMSE and correlation with BDNF. The repeated measures ANOVA for the
longitudinal MMSE data (Figure 2; Appendix, Table 2) showed time and BDNF group had a significant effect on MMSE scores, but there was no significant interaction between time and BDNF group. Meaning that MMSE scores changed significantly over time after ECT in the whole group and that there was a difference in MMSE scores between the BDNF groups, but the BDNF groups did not differ in the way the MMSE scores changed over time. The repeated measures ANOVA also showed
educational level, age and psychotic symptoms were significantly correlated with MMSE scores. Post-hoc analysis with a repeated measures ANOVA showed the low and high BDNF group only differed significantly in MMSE scores 6 months after ECT (Figure 2). The other repeated measures ANOVA showed the MMSE scores changed significantly over time after ECT in both the low (F(1, 223) = 16.1,
p < .001) and high (F(1, 202) = 7.27, p = .008) BDNF group. With pairwise t-tests it was found that in
the low BDNF group, baseline MMSE scores were significantly different from the MMSE scores 1 week (p = .01), 4 weeks (p < .001) and 6 months (p < .001) after ECT. The scores after the sixth ECT were also significantly different from the scores 1 week (p = .04), 4 weeks (p < .001) and 6 months (p < .001) after ECT. MMSE scores from 1 week after ECT differed significantly from the scores 6 months after ECT (p = .02) and lastly, the scores 6 months after ECT differed significantly from 5 years after ECT (p = .01). For the high BDNF groupbaseline MMSE was significantly different from the scores 4 weeks (p < .001) and 6 months (p = .04) after ECT. The scores after the sixth ECT differed significantly from those 4 weeks after the last ECT (p < .001) as did the scores from 1 week after ECT (p = .03)
10 Table 1
Sociodemographic Characteristics of Participants at Baseline
Baseline characteristic Total (N = 86) Low BDNFa (N = 43) High BDNFa (N = 43) χ2/t/U (df) p- value (Low vs. High) Age mean (SD) 72.9 (8.4) 73.3 (8.2) 72.5 (8.6) 0.44 (84) .66 Gender (female) 57 (64.8) 27 (62.8) 30 (69.8) 0.47 (1) .49 Level of education N = 73 - Low - Middle - High 40 (54.8) 20 (27.4) 13 (17.8) 22 (56.4) 11 (28.2) 6 (15.4) 18 (52.9) 9 (26.5) 7 (20.6) 0.34 (2) .85 Site (Amsterdam) 43 (50.0) 13 (30.2) 30 (69.8) 13.44 (1) < .001* Late onset depression (>55 years) 49 (57.0) 26 (60.5) 23 (53.5) 0.43 (1) .51 Depression with psychosis 45 (52.3) 26 (60.5) 19 (44.2) 2.28 (1) .13 Physical comorbidity - Any present - Cardiovascular disease - Hypertension 70 (81.4) 21 (24.4) 24 (27.9) 30 (69.8) 10 (23.3) 9 (20.9) 40 (93.0) 11 (25.6) 15 (34.9) 6.22 (1) .01* 0.06 (1) .80 2.08 (1) .15 Treatments during current episode
- Antidepressants - Mood stabilizers - Antipsychotics - Benzodiazepine - Other 73 (84.9) 23 (26.7) 27 (31.4) 35 (40.7) 10 (11.6) 34 (79.1) 5 (11.6) 13 (30.2) 14 (32.6) 6 (14.0) 39 (90.7) 18 (41.9) 14 (32.6) 21 (48.8) 4 (9.3) 2.27 (1) .13 10.03 (1) .002* 0.05 (1) .82 2.36 (1) .12 0.11 (1) .74 Duration of index episode
(months) mean (SD) 12.5 (20.4) 11.1 (23.2) 13.8 (17.4) -1.42 (77) .16b Alcohol N = 82 - Never - Sometimes - Regularly 52 (63.4) 11 (13.4) 19 (23.2) 24 (57.1) 6 (14.3) 12 (28.6) 28 (70.0) 5 (12.5) 7 (17.5) 1.67 (2) .43 Smoking N = 72 - Never - Currently - Ever 47 (65.3) 18 (25.0) 7 (9.7) 24 (72.7) 6 (18.2) 3 (9.1) 23 (59.0) 12 (30.8) 4 (10.3) 1.68 (2) .43
Clinical status 1 week after ECT - Response - Remission 69 (80.2) 59 (68.6) 38 (88.4) 35 (81.4) 31 (72.1) 24 (55.8) 3.59 (1) .06 6.53 (1) .01* MADRS mean (SD) - Pre-ECT - Post-ECT 33.8 (9.2) 9.7 (9.5) 33.8 (8.9) 8.3 (9.9) 33.8 (9.6) 11.2 (9.0) 0.03 (81) 0.98 -1.86 (84) 0.07b BDNF (ng/ml) pre-ECT median (IQR) 17.9 (7.6) 13.8 (7.9) 22.7 (3.7) 0 (84) < .001*c
MMSE pre-ECT 24.1 (5.2) 24.5 (5.8) 23.6 (4.5) -0.30 (76) 0.77d
Note. Data shown as n (%), unless indicated otherwise. In case of missing data for certain variables, the number of complete
cases (n=) is presented in the left column. Education: low (no education, primary school), middle (high school, vocational training), high (college, university). Treatments: antidepressants (Selective Serotonin Reuptake Inhibitors, Tricyclic antidepressants, Monoamine-oxidase inhibitors), mood stabilizers (lithium, other mood stabilizers) Alcohol: sometimes (maximum of 4 times a month), regularly (at least 2 times a week). Brain Derived Neurotrophic Factor (BDNF),
Electroconvulsive Therapy (ECT), Montgomery Asberg Depression Rating Scale (MADRS), Mini Mental State Examination (MMSE), Interquartile range (IQR), nanograms (ng), millilitres (ml).
a BDNF levels were split with median (17.9 ng/ml) as cut-off, resulting in low and high BDNF group b Skewed data, t-tests performed on log-transformed data.
c Mann-Whitney U test performed with caution because of different types of skew. d Skewed data, t-tests performed on reverse log-transformed data
11 Figure 1
Means and Standard Errors of BDNF Levels Over Time in Low and High BDNF Group and Test Statistics of Repeated measures ANOVA Comparing BDNF Levels of Low and High BDNF Group at Each Time Point
Note. Error bars represent standard errors. Brain Derived Neurotrophic Factor (BDNF), Electroconvulsive therapy (ECT),
Mean (M), Standard Deviation (SD), Degrees of Freedom Numerator (dfn), Degrees of Freedom (dfd) *p < .05
12 Figure 2
Means and Standard Errors of MMSE Scores Over Time in Low and High BDNF Group and Test Statistics of Repeated measures ANOVA Comparing MMSE Scores of Low and High BDNF Group at Each Time Point
Note. Means and standard errors of raw data are displayed, statistical analysis was performed on reversed, log transformed
MMSE scores. Error bars represent standard errors. Mini Mental State Examination (MMSE), Brain Derived Neurotrophic Factor (BDNF), Electroconvulsive therapy (ECT), Mean (M), Standard Deviation (SD), Degrees of Freedom Numerator (dfn), Degrees of Freedom (dfd)
13 Discussion
In this study the longitudinal changes of BDNF levels and MMSE scores and the association between baseline BDNF and longitudinal MMSE scores were investigated in severe LLD patients who were treated with ECT. The aim was to gain knowledge about the underlying mechanism of the changes in global cognitive functioning after ECT in this patient group.
It was found that patients with low and high BDNF levels at baseline differed in their trajectory of BDNF levels after ECT. BDNF levels only increased over time in patients with low BDNF levels at baseline. Nevertheless, BDNF levels remained significantly higher in patients from the high BDNF group at all time points. Furthermore, MMSE scores improved until 6 months after ECT in patients with low BDNF levels at baseline and in the high BDNF group the improvement of MMSE scores was significant until 4 weeks after ECT. Even though an effect of BDNF group on MMSE scores was found, there was no association between baseline BDNF levels and baseline MMSE scores and patients with low and high BDNF levels at baseline did not differ in their trajectory of MMSE scores after ECT. There was only a difference in MMSE scores between the two BDNF groups 6 months after ECT, such that MMSE scores were higher in the low BDNF group.
Although previous studies found increased BDNF levels after ECT in their whole sample (Bocchio-Chiavetto et al., 2006; Brunoni, Baeken, Machado-Vieira, Gattaz & Vanderhasselt, 2014), this study only found increased BDNF levels in the low BDNF group. However, only Bocchio-Chiavetto et al. (2006) made a distinction between a low and high BDNF group and they also found a difference between these groups in their trajectory of BDNF levels. In the high BDNF group BDNF levels only changed significantly after ECT, whereas BDNF levels in the low BDNF group also changed
significantly during ECT. Possibly, the BDNF levels of patients with high BDNF levels never decreased during their depression and have always remained ‘normal’. In healthy, elderly subjects an average of BDNF level of 23.3 ng/ml was found (Ziegenhorn et al., 2007), which is close to the mean at baseline in the high BDNF group in this study, 22.9 ng/ml. Consequently, BDNF might not increase in these patients since it is already close to ‘normal’ levels. Perhaps, the absence of an increase is related to the finding that the high BDNF group included less remitted patients. This is in line with van Zutphen et al. (2019) who found that lower BDNF levels increased the chance of remission. Possibly, an increase of BDNF is needed for remission.
High BDNF levels at baseline were also associated with the use of mood stabilizers, which could also be the cause of the high BDNF levels. This is substantiated by a study that found an increase of BDNF after the use of lithium in patients with bipolar disorder. However, little research has been done about the effects of mood stabilizers on BDNF, so more evidence is needed to confirm this. Although
14 another characteristic of the high BDNF group was the presence of any physical comorbidity, it is unclear how this characteristic is related to BDNF levels. Previously, BDNF has mainly been associated with the physical comorbidities cardiovascular disease and hypertension (Pius-Sadowska &
Machaliński, 2017), but this was not verified in the current study.
MMSE scores improved after ECT in all patients, regardless of BDNF levels at baseline, which was in accordance with previous studies (Hausner et al., 2011; Obbels et al., 2019). However, in patients with low BDNF levels at baseline, MMSE scores declined again between 6 months and 5 years after ECT, at which point they did not significantly differ from baseline anymore. In the high BDNF group, a non-significant decline started from 4 weeks after ECT, causing a non-significant difference in MMSE scores 5 years after ECT compared to baseline. Maybe the less persistent improvement of MMSE scores in patients with higher BDNF levels is explained by the relation of BDNF with neuronal plasticity (Bus & Molendijk, 2016). Since BDNF levels are already high, neuronal plasticity is also higher, which leaves less room for improvement. Even though previous studies did not investigate the relation between BDNF and long-term cognition after ECT, two other studies found concurrent results. They both found a significant improvement in cognition 4 years after ECT in patients with LLD and cognitive impairment and a non-significant decrease compared to 15 months after ECT
(Stoudemire, Hill, Morris & Dalton, 1995; Stoudemire, Hill, Morris, Martino-Saltzman & Lewison, 1993). Thus, based on existing studies cognitive improvement is probably maintained up to 15 months after ECT, but can decrease again 4 or 5 years after ECT. Possibly, because the effect of ECT no longer persists or it could be an effect of age, which was negatively correlated with MMSE scores in this study.
The hypothesis that BDNF levels and MMSE scores are positively related during and after ECT was not substantiated in this study. There was no association between BDNF levels and longitudinal MMSE scores. Furthermore, MMSE scores were higher in patients with high BDNF levels, although the difference was not significant at most time points (Figure 2). This would indicate a negative relationship between BDNF levels and MMSE scores, opposite of what has been found in previous studies (Engelmann et al., 2019; Gunstad et al., 2008; Laske et al., 2011). However, those studies were performed either in elderly without depression or in adults with depression. In addition, Dols et al. (2015) did not find a relation between BDNF and cognition in LLD. Combined with the results of this study, this could indicate that BDNF works differently in severe depression in the elderly and that there is another pathophysiological mechanism underlying the cognitive changes after ECT. It is also possible that the relation between BDNF and cognition is distinct for a specific cognitive domain. This study used the MMSE, which is a measure for global cognitive functioning, but it would be likely that BDNF is mainly related to memory functions. That is because BDNF is mostly found in the
15 hippocampus, which plays a big role in memory (Duman & Monteggia, 2006). And indeed, previous research found a positive association between BDNF and hippocampal volume in LLD patients (Bouckaert et al., 2016) and showed a relation between low BDNF levels, smaller hippocampi and poorer memory (Erickson et al., 2010).
The strengths of this study include the relatively large sample size for such a specific patient group. In addition, statistical analyses were adjusted for multiple variables, including age and educational level, which both had a significant effect on MMSE scores. A limitation of the study is the lack of a control group to measure the longitudinal changes of BDNF levels and cognition without the
intervention of ECT. Ideally this control group would consist of severely depressed elderly who use as little medication as possible, however, this is hard to set up, since adequate treatment should always be first priority. Another reason for caution in interpreting the results is the method of measuring BDNF. Serum BDNF is a peripheral measure and not directly related to BDNF levels in the brain. Even though BDNF can cross the blood-brain barrier, it is unclear how serum BDNF levels reflect cerebral BDNF, since BDNF is also produced in the rest of the body (Gass & Hellweg, 2010).On the other hand, measuring serum BDNF is very common in studies that examine BDNF levels in humans, so results are comparable.
The results of this study can be used in future research. First of all, LLD patients with low and high BDNF levels at baseline differed in their trajectory of BDNF levels and in remission rates. Maybe, the lack of BDNF increase in patients with high BDNF levels at baseline is related to a lower chance of remission. Future reports could use the longitudinal BDNF data and study its relation with remission after ECT. Plus, the characteristics of patients without increasing BDNF levels could be studied, to learn more about possible reasons why some patients do not go into remission. Moreover, this study did not find a relation between baseline BDNF levels and MMSE scores before, during and after ECT. However, the relation between longitudinal changes in BDNF levels and longitudinal MMSE was not studied and perhaps the changes in BDNF levels can explain the changes in MMSE scores induced by ECT. For example, patients with high BDNF levels showed an earlier decline in MMSE scores, which could be related to the lack of increase in BDNF. Lastly, as discussed earlier, this study measured global cognitive functioning, but it is possible that BDNF levels are mainly related to memory instead of cognition in general. Thus, future research should investigate the relation between BDNF levels and memory in LLD patients who are treated with ECT. To pursue this, results from
neuropsychological tests included in the MODECT dataset can be used. These include the Rey Auditory Verbal Learning Test, the Visual Association Test and the Category Fluency Test that assess verbal, visual and semantic memory, respectively.
16 The current findings are important, since they can take away concerns that have been raised about the safety of ECT and its impact on cognition, especially for elderly. This study shows that ECT did not cause cognitive impairment in severe LLD patients and that cognition can be improved, even in people with low BDNF levels. In addition, this study showed a possible characteristic of severe LLD patients who do not go in remission, namely, high BDNF levels. Moreover, even though no
correlation was found between BDNF levels and global cognitive functioning, this study added to the knowledge about ECT and its effects in severe LLD patients.
In conclusion, serum BDNF levels only increased after ECT in severe LLD patients with low BDNF levels at baseline and irrespective of BDNF levels, cognition improved up to 6 months after ECT. So far, no relation was found between baseline BDNF levels and cognition before, during and after ECT. Thus, the underlying mechanism of cognitive changes after ECT in severe LLD patients remains unclear.
17 References
Adler, G., Chwalek, K., & Jajcevic, A. (2004). Six-month course of mild cognitive impairment and affective symptoms in late-life depression. European Psychiatry, 19(8), 502–505.
https://doi.org/10.1016/j.eurpsy.2004.09.003
American Psychiatric Association. (2000). Diagnostic and Statistical Manual of Mental Disorders (4th, text ed.). Washington DC: American Psychiatric Press.
Birkenhager, T. K., de Boer, D., Burggraaf, J. P., & van den Broek, W. W. (2010). Richtlijn
elektroconvulsietherapie [Guidelines Electroconvulsive Therapy] (1st ed.). De Tijdstroom.
Bocchio-Chiavetto, L., Zanardini, R., Bortolomasi, M., Abate, M., Segala, M., Giacopuzzi, M., Riva, M. A., Marchina, E., Pasqualetti, P., Perez, J., & Gennarelli, M. (2006). Electroconvulsive therapy (ECT) increases serum brain derived neurotrophic factor (BDNF) in drug resistant depressed patients. European Neuropsychopharmacology, 16(8), 620–624.
https://doi.org/10.1016/j.euroneuro.2006.04.010
Bouckaert, F., Dols, A., Emsell, L., De Winter, F. L., Vansteelandt, K., Claes, L., Sunaert, S., Stek, M., Sienaert, P., & Vandenbulcke, M. (2016). Relationship between hippocampal volume, serum BDNF, and depression severity following electroconvulsive therapy in late-life depression.
Neuropsychopharmacology, 41(11), 2741–2748. https://doi.org/10.1038/npp.2016.86
Brunoni, A. R., Baeken, C., Machado-Vieira, R., Gattaz, W. F., & Vanderhasselt, M. A. (2014). BDNF blood levels after electroconvulsive therapy in patients with mood disorders: A systematic review and meta-analysis. World Journal of Biological Psychiatry, 15(5), 411–418.
https://doi.org/10.3109/15622975.2014.892633
Bus, B. A.A., & Molendijk, M. L. (2016). De neurotrofe hypothese van depressie [The neurotrophic hypothesis of depression]. Tijdschrift Voor Psychiatrie, 58(3), 215–222.
Bus, B. A.A., Molendijk, M. L., Tendolkar, I., Penninx, B. W. J. H., Prickaerts, J., Elzinga, B. M., & Voshaar, R. C. O. (2015). Chronic depression is associated with a pronounced decrease in serum brain-derived neurotrophic factor over time. Molecular Psychiatry, 20(5), 602–608.
https://doi.org/10.1038/mp.2014.83
Bus, Boudewijn A.A., Tendolkar, I., Franke, B., de Graaf, J., Heijer, M. Den, Buitelaar, J. K., & Oude Voshaar, R. C. (2012). Serum brain-derived neurotrophic factor: Determinants and relationship with depressive symptoms in a community population of middle-aged and elderly people. The
World Journal of Biological Psychiatry, 13(1), 39–47.
https://doi.org/10.3109/15622975.2010.545187
Dols, A., Bouckaert, F., Sienaert, P., Rhebergen, D., Vansteelandt, K., ten Kate, M., de Winter, F. L., Comijs, H. C., Emsell, L., Oudega, M. L., van Exel, E., Schouws, S., Obbels, J., Wattjes, M., Barkhof, F., Eikelenboom, P., Vandenbulcke, M., & Stek, M. L. (2017). Early- and Late-Onset Depression in Late Life: A Prospective Study on Clinical and Structural Brain Characteristics and Response to Electroconvulsive Therapy. American Journal of Geriatric Psychiatry, 25(2), 178– 189. https://doi.org/10.1016/j.jagp.2016.09.005
Dols, A., Thesing, C. S., Bouckaert, F., Voshaar, R. C. O., Comijs, H. C., & Stek, M. L. (2015). BDNF serum levels are not related to cognitive functioning in older depressed patients and controls.
International Psychogeriatrics, 27(4), 649–656. https://doi.org/10.1017/S1041610214002622
Duman, R. S., & Monteggia, L. M. (2006). A neurotrophic model for stress-related mood disorders.
Biological Psychiatry, 59(12), 1116–1127. https://doi.org/10.1016/j.biopsych.2006.02.013
18 Müller, M. B., Tüscher, O., Frieling, H., Tadić, A., & Lieb, K. (2019). Higher BDNF plasma levels are associated with a normalization of memory dysfunctions during an antidepressant treatment. European Archives of Psychiatry and Clinical Neuroscience, 270(2), 183–193. https://doi.org/10.1007/s00406-019-01006-z
Erickson, K. I., Prakash, R. S., Voss, M. W., Chaddock, L., Heo, S., McLaren, M., Pence, B. D., Martin, S. A., Vieira, V. J., Woods, J. A., McAuley, E., & Kramer, A. F. (2010). Brain-derived neurotrophic factor is associated with age-related decline in hippocampal volume. Journal of Neuroscience,
30(15), 5368–5375. https://doi.org/10.1523/JNEUROSCI.6251-09.2010
Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12(3), 189–198. https://doi.org/10.1016/0022-3956(75)90026-6
Gass, P., & Hellweg, R. (2010). Peripheral brain-derived neurotrophic factor (BDNF) as a biomarker for affective disorders? The International Journal of Neuropsychopharmacology, 13(01), 1. https://doi.org/10.1017/S1461145709991039
Gunstad, J., Spitznagel, M. B., Glickman, E., Alexander, T., Juvancic-Heltzel, J., Walter, K., & Murray, L. (2008). β-Amyloid is associated with reduced cognitive function in healthy older adults. Journal
of Neuropsychiatry and Clinical Neurosciences, 20(3), 327–330.
https://doi.org/10.1176/jnp.2008.20.3.327
Hausner, L., Damian, M., Sartorius, A., & Frölich, L. (2011). Efficacy and cognitive side effects of electroconvulsive therapy (ECT) in depressed elderly inpatients with coexisting mild cognitive impairment or dementia. Journal of Clinical Psychiatry, 72(1), 91–97.
https://doi.org/10.4088/JCP.10m05973gry
Korten, N. C. M., Penninx, B. W. J. H., Kok, R. M., Stek, M. L., Oude Voshaar, R. C., Deeg, D. J. H., & Comijs, H. C. (2014). Heterogeneity of late-life depression: Relationship with cognitive functioning. International Psychogeriatrics, 26(6), 953–963.
https://doi.org/10.1017/S1041610214000155
Laske, C., Stellos, K., Hoffmann, N., Stransky, E., Straten, G., Eschweiler, G. W., & Leyhe, T. (2011). Higher BDNF serum levels predict slower cognitive decline in Alzheimer’s disease patients. The
International Journal of Neuropsychopharmacology, 14(03), 399–404.
https://doi.org/10.1017/S1461145710001008
Lee, J. S., Potter, G. G., Wagner, H. R., Welsh-Bohmer, K. A., & Steffens, D. C. (2007). Persistent mild cognitive impairment in geriatric depression. International Psychogeriatrics, 19(1), 125–135. https://doi.org/10.1017/S1041610206003607
Molendijk, M. L., Spinhoven, P., Polak, M., Bus, B. A. A., Penninx, B. W. J. H., & Elzinga, B. M. (2014). Serum BDNF concentrations as peripheral manifestations of depression: Evidence from a systematic review and meta-analyses on 179 associations (N=9484). Molecular Psychiatry,
19(7), 791–800. https://doi.org/10.1038/mp.2013.105
Montgomery, S. A., & Asberg, M. (1979). A new depression scale designed to be sensitive to change.
British Journal of Psychiatry, 134(4), 382–389. https://doi.org/10.1192/bjp.134.4.382
Naismith, S. L., Norrie, L. M., Mowszowski, L., & Hickie, I. B. (2012). The neurobiology of depression in later-life: Clinical, neuropsychological, neuroimaging and pathophysiological features. Progress
in Neurobiology, 98(1), 99–143. https://doi.org/10.1016/j.pneurobio.2012.05.009
Obbels, J., Vansteelandt, K., Verwijk, E., Dols, A., Bouckaert, F., Oudega, M. L., Vandenbulcke, M., Stek, M., & Sienaert, P. (2019). MMSE changes during and after ECT in late-life depression: A
19 prospective study. American Journal of Geriatric Psychiatry, 27(9), 934–944.
https://doi.org/10.1016/j.jagp.2019.04.006
Pius-Sadowska, E., & Machaliński, B. (2017). BDNF – A key player in cardiovascular system. Journal of
Molecular and Cellular Cardiology, 110, 54–60. https://doi.org/10.1016/j.yjmcc.2017.07.007
Shéehan, D. V, Lecrubier, Y., Sheehan, K. H., Amorim, P., Janavs, J., Weller, E., Hergueta, T., Baker, R., & Dubar, G. C. (1998). The Mini-International Neuropsychiatric Interview (M.I.N.I): The
development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. The Journal of Clinical Psychiatry, 59(Suppl 20), 22–33.
Spaans, H. P., Sienaert, P., Bouckaert, F., Van Den Berg, J. F., Verwijk, E., Kho, K. H., Stek, M. L., & Kok, R. M. (2015). Speed of remission in elderly patients with depression: Electroconvulsive therapy V. medication. British Journal of Psychiatry, 206(1), 67–71.
https://doi.org/10.1192/bjp.bp.114.148213
Stoudemire, A., Hill, C. D., Morris, R., & Dalton, S. T. (1995). Improvement in depression-related cognitive dysfunction following ECT. The Journal of Neuropsychiatry and Clinical Neurosciences,
7(1), 31–34. https://doi.org/10.1176/jnp.7.1.31
Stoudemire, A., Hill, C. D., Morris, R., Martino-Saltzman, D., & Lewison, B. (1993). Long-term affective and cognitive outcome in depressed older adults. American Journal of Psychiatry, 150(6), 896– 900. https://doi.org/10.1176/ajp.150.6.896
van Zutphen, E. M., Rhebergen, D., van Exel, E., Oudega, M. L., Bouckaert, F., Sienaert, P.,
Vandenbulcke, M., Stek, M., & Dols, A. (2019). Brain-derived neurotrophic factor as a possible predictor of electroconvulsive therapy outcome. Translational Psychiatry, 9(1).
https://doi.org/10.1038/s41398-019-0491-9
Volkert, J., Schulz, H., Härter, M., Wlodarczyk, O., & Andreas, S. (2013). The prevalence of mental disorders in older people in Western countries - a meta-analysis. Ageing Research Reviews,
12(1), 339–353. https://doi.org/10.1016/j.arr.2012.09.004
Ziegenhorn, A. A., Schulte-Herbrüggen, O., Danker-Hopfe, H., Malbranc, M., Hartung, H. D., Anders, D., Lang, U. E., Steinhagen-Thiessen, E., Schaub, R. T., & Hellweg, R. (2007). Serum
neurotrophins-A study on the time course and influencing factors in a large old age sample.
Neurobiology of Aging, 28(9), 1436–1445.
20 Appendix
Table 1
Test Statistics of Repeated Measures ANOVA with Longitudinal Data of BDNF Levels
Variable SS dfn, dfd F p η2
Age 86.28 1, 300 4.10 .04* .01
Gender 2.82 1, 300 0.13 .71 < .001
Inclusion site 628.40 1, 300 29.86 < .001* .09
Physical comorbidity 31.58 1, 300 1.50 .22 .005
Remission after ECT Use of mood stabilizers
36.36 0.119 1, 300 1, 300 1.73 0.01 .19 .94 .006 < .001 BDNF group 2598.21 1, 300 123.48 < .001* .29 Time 63.30 3, 300 1.00 .39 .01 BDNF group x Time 251.56 3, 300 3.99 .008* .04
Note. Sums of Squares (SS), Degrees of Freedom Numerator (dfn), Degrees of Freedom Denominator (dfd), Brain Derived
Neurotrophic Factor (BDNF). * p < .05
Table 2
Test Statistics of Repeated Measures ANOVA with Longitudinal Data of MMSE Scores
Variable SS dfn, dfd F p η2 Age 4.21 1, 357 10.04 .002* .03 Gender 0.33 1, 357 0.79 .38 .002 Educational level 14.23 2, 357 16.98 < .001* .09 Presence of psychotic symptoms 2.07 1, 357 4.94 .03* .01 BDNF group 3.47 1, 357 8.29 .004* .02 Time 28.43 5, 357 13.57 < .001* .16 BDNF group x Time 1.07 5, 357 0.51 .77 .01
Note. Statistical analysis was performed on log transformed, reversed MMSE scores. Sums of Squares (SS), Degrees of
Freedom Numerator (dfn), Degrees of Freedom Denominator (dfd), Brain Derived Neurotrophic Factor (BDNF). * p < .05
