Abstract
This study: 1) examined cross-sectional and longitudinal relations of serum brain-derived neurotrophic factor (BDNF) to late-life depression (LLD); 2) tested effects of vitamin D3 and omega-3s on change in BDNF; 3) explored modifying or mediating roles of BDNF on effects of vitamin D3 and omega-3s for LLD. We selected 400 adults from a completed trial of vitamin D3 and omega-3 supplements for LLD prevention. BDNF was measured using an enzyme-linked immunosorbent assay. We administered semi-structured diagnostic interviews and Patient Health Questionnaire [PHQ]-9 to ascertain outcomes at baseline (depression caseness vs. non-caseness; PHQ-9) and at 2-year follow-up among baseline non-depressed individuals (incident vs. no incident MDD; change in PHQ-9). At baseline, while there were no significant differences in mean serum BDNF comparing depression cases and non-cases, being in the lowest vs. highest serum BDNF quartile was significantly associated with worse depressive symptoms. There were no significant longitudinal associations between serum BDNF and LLD. Neither supplement significantly affected change in BDNF; serum BDNF did not appear to modify or mediate treatment effects on LLD. In conclusion, we observed significant cross-sectional but not longitudinal associations between serum BDNF levels and LLD. Vitamin D3 or omega-3s did not alter serum BDNF over 2 years.
Keywords: BDNF, depression, vitamin D3, omega-3s, geriatric
1. INTRODUCTION
Major depressive disorder (MDD) affects millions worldwide and is a major cause of later-life disability. Clinical heterogeneity and biological complexity of MDD pose significant challenges in discovering reliable clinical biomarkers (Strawbridge et al., 2017). Furthermore, the identification of minimally-invasive, cost-effective, and easy-to-obtain biomarkers can serve as potential targets for detection, treatment, and prevention of late-life depression (LLD). Brain-derived neurotrophic factor (BDNF), a member of the neurotrophin family of growth factors, is synthesized in the brain, can cross the blood-brain barrier, and is abundantly distributed throughout the peripheral tissues. BDNF plays a pivotal role in neuronal differentiation, maintenance, and survival, as well as in the synaptic plasticity of neurons – all biological processes implicated in the development of MDD (Bathina and Das, 2015; Duman et al., 2000; Duman and Monteggia, 2006). Thus, low serum levels of BDNF have promise as a biomarker in MDD. In meta-analyses of observational studies, depressed patients had lower serum BDNF compared to non-depressed individuals; however, clinical heterogeneity of participants and study designs limit the ability to draw uniform conclusions across these studies (Bocchio-Chiavetto et al., 2010; Molendijk et al., 2014; Shi et al., 2020). Furthermore, little is known about whether serum BDNF can predict new onset of MDD and change in mood in older adults.
The role of nutraceuticals, such as vitamin D3 and marine omega-3 fatty acids (omega-3s) supplements, have been previously investigated for prevention of LLD in randomized clinical trials (RCTs); the majority of RCTs found no benefits of either supplement on LLD prevention (Liao et al., 2019; Vellekkatt and Menon, 2019). Experimental rat model studies found that supplementation with vitamin D3 may increase serum BDNF levels by activating BDNF gene expression pathways (Khairy and Attia, 2021). Additionally, prior animal studies observed that vitamin D3 may indirectly modulate apoptosis and age-related pro-inflammatory states via stimulation of BDNF production (Briones and Darwish, 2012; Nagatsu and Sawada, 2005). Similarly, evidence suggests that dietary omega-3s enhance BDNF synthesis by activating BDNF transcription through PI3K/Akt signaling pathway and thereby increasing BDNF levels (Knöchel et al., 2015; Wu et al., 2008). However, limited trial evidence exists on whether vitamin D3 and omega-3s can alter serum BDNF in a well-characterized sample of older adults (Gravesteijn et al., 2022). Also, knowledge gaps remain regarding serum BDNF as a biologic modifier or mediator, which could inform how vitamin D3 and omega-3s influence LLD risk and for whom (Papakostas and Fava, 2008).
Thus, we addressed these knowledge gaps among an in-clinic sample of 400 participants in VITAL-DEP (VITamin D and OmegA-3 TriaL-Depression Endpoint Prevention) study (Okereke et al., 2018). Our study objectives were to: 1) examine cross-sectional and prospective relationships between serum BDNF and LLD; 2) test effects of vitamin D3 and/or omega-3s on change in serum BDNF over 2 years; 3) explore potential moderating or mediating roles of BDNF on the effects of vitamin D3 or omega-3s for LLD prevention.
2. METHODS
2.1. Source of participants
Participants were members of the VITAL-DEP (NCT01696435), depression prevention ancillary study to VITAL (NCT01169259). VITAL tested vitamin D3 (2000 IU/day) and/or omega-3s (1 g/day) in a 2 x 2 factorial design for primary prevention of cardiovascular disease and cancer in 25,871 US adults (men aged ≥50 and women aged ≥55 years); dosing and constituents of the study agents in VITAL-DEP were constrained to being the same as in the parent trial. The detailed protocols of VITAL and VITAL-DEP trials are described elsewhere (Manson et al., 2012; Okereke et al., 2018). A sub-cohort of 1054 VITAL participants from the New England region were enrolled for detailed, in-person phenotyping at baseline (period: January 2012 and March 2014) and 2 years later (period: January 2014 – April 2016) at the Clinical and Translational Science Center (CTSC) in Boston, MA, USA. VITAL participants also provided fasting blood samples during CTSC visits. All participants provided written informed consent, and approvals were obtained from the institutional review board of Mass General Brigham.
All 1,054 VITAL-CTSC participants were invited to take part in the 45-minute VITAL-DEP assessment, which featured a semi-structured psychiatric diagnostic interview, neuropsychological testing, and self-report questionnaires on mood and other health factors. The purpose of this assessment was to identify individuals at risk for incident MDD at a 2-year follow-up (Okereke et al., 2018). Of 1,046 participants who completed the baseline assessment, 326 were ineligible for a 2-year follow-up due to a baseline history of MDD and/or other major psychiatric exclusions (Figure S1). Of the remaining 720 participants, 662 (91.9%) completed follow-up MDD assessments.
2.2. Sample selection
Among 326 participants with a history of depressive and/or other major psychiatric disorders and with a CTSC blood sample at baseline, we randomly selected 100 participants with a history of depression for baseline serum BDNF assay; the sample was balanced by sex and randomized treatment agents. The depression cases included current, past, or recurrent MDD, current or past major depressive episode, or dysthymia.
Among participants who were free of any major psychiatric disorders at baseline, who completed 2-year follow-up assessments, and who provided blood samples at both baseline and 2-year follow-up visits, we selected 300 participants for baseline and year-2 serum BDNF assays. For this selection, we included all 34 incident MDD cases (identified during the 2-year VITAL-DEP follow-up assessments) to increase the power for longitudinal analysis; the other 266 participants did not have incident MDD over 2 years and were balanced by sex and treatment agents (See Figure S1).
2.3. Assessment of depression outcomes
The Mini-International Neuropsychiatric Interview (MINI) (Sheehan et al., 1998) for the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV was used to ascertain MDD and psychiatric diagnosis status at baseline and 2-year follow-up. Participants also completed self-report measures (including Patient Health Questionnaire [PHQ]-9; range 0-27 points) (Kroenke et al., 2001) at baseline and 2-year CTSC visits. In the main VITAL trial, participants also completed the PHQ-8 (range 0-24 points) (Kroenke et al., 2009) at baseline and annually thereafter over a median 5 years of follow-up.
2.4. Serum BDNF assays
Samples were thawed, aliquoted, and sent to the Rifai laboratory at Boston Children’s Hospital for serum BDNF assay measured by a commercially available enzyme-linked immunosorbent assay kit (Quantikine™) (Crombie et al., 2021); assays were performed according to the manufacturer’s instructions. To minimize batch effects, all baseline and year 2 sample pairs were shipped and measured at the same time in a manner blinded to the analyzing lab. Among 36 blinded QC replicates randomly distributed across the plates, the average coefficient of variation (CV) was 5.8%; within-pair CV was 3.1%.
2.5. Statistical analyses
2.5.1. Samples for cross-sectional and longitudinal analyses
To mitigate risk of selection bias, we selected at random a sub-set of 266 participants who had no major psychiatric disorders at baseline and did not develop MDD over 2 years, along with a random selection of 14 of the 34 participants who had no history of major psychiatric disorders at baseline but who developed incident MDD over 2 years. The random selection of 14 of the 34 participants was performed to match the MDD incidence rate observed in the overall VITALDEP CTSC sub-cohort (i.e., 34/720*300) and to preserve the caseness in the baseline sample. Thus, we performed the baseline cross-sectional analyses in a randomly selected sample of 100 participants with a history of depression and 280 (266 + 14) participants without depression or other major psychiatric history at baseline.
As described in sample selection, we selected 300 participants for baseline and year-2 serum BDNF assays. Of 300 participants, we included n=288 for longitudinal analyses. We excluded prevalent selective serotonin reuptake inhibitor (SSRI) users (n=10) or those with missing self-reported information on SSRI use (n=2) at baseline, as SSRIs are frequently prescribed for depression and other psychiatric disorders (Edinoff et al., 2021), and evidence suggests that SSRIs increase BDNF levels by enhancing BDNF gene expression (Björkholm and Monteggia, 2016; Zhou et al., 2017). Thus, we excluded participants who were prevalent SSRI users or those with missing information on SSRI use to mitigate potential bias in the estimates for longitudinal analyses.
2.5.2. Descriptive analyses
We examined the distribution of baseline serum BDNF using a box plot and histogram and found this variable to be normally distributed. In the sample eligible for cross-sectional analyses (n=380), we compared participants’ characteristics according to depression status using two-sample t-tests for normally distributed continuous variables, Wilcoxon’s rank-sum tests for non-normally distributed continuous variables, and chi-square tests for difference in proportions.
We also compared baseline characteristics according to quartiles of baseline serum BDNF in the sample eligible for longitudinal analyses (n=288); p-values for trend were calculated using: Jonckheere-Terpstra test for non-parametric continuous variables; Cochran-Armitage trend test for binary categorical variable; Cochran Mantel-Haenszel statistics for a categorical variable with more than 2 levels.
2.5.3. Cross-sectional analyses
We constructed multivariable logistic regression models to examine the cross-sectional associations between serum BDNF level and depression status (i.e., participants with a history of depression vs. those without any history of major psychiatric disorders). Furthermore, we used multivariable negative binomial models for examining the cross-sectional associations between quartiles of serum BDNF and PHQ-9 scores; percent differences and 95% confidence intervals (CIs) were reported. The choice of covariates for inclusion in multivariable models was based on prior evidence (such as age, SSRI use) and on evidence of p<0.15 on stepwise covariate selection (such as sex, Black vs. non-Black, and history of diabetes) (Bursae et al., 2008); see supplement for more details on stepwise approach.
2.5.4. Longitudinal analyses
We used multivariable logistic regression models to determine odds ratio (OR) and 95% CI of incident MDD over 2 years by a median of baseline serum BDNF. After using measures for goodness of fit and Voung test (Akaike, 1973; Voung, 1989), we determined that the negative binomial model, compared to the zero-inflated negative binomial model, was most appropriate to the PHQ-9 distribution; thus, we used repeated measures negative binomial regression to estimate the association between baseline serum BDNF level and percent differences in change in PHQ-9 over 2 years; time was modeled as an indicator variable, and the p-interaction was reported using time-x-quartiles of serum BDNF. All longitudinal models were adjusted for age, sex, and treatment agents.
2.5.5. Testing treatment agents’ effects on serum BDNF
We examined effects of vitamin D3 or omega-3s, vs. placebos, on 2-year change in serum BDNF using repeated measures response profile models; the mean difference in change in serum BDNF between treatment vs. placebo groups was assessed using a time-x-treatment interaction.
2.5.6. Exploratory analyses of modification and mediation
We probed effect modification by baseline serum BDNF (i.e., ≥ or <median level) of the effects of vitamin D3 or omega-3s, vs. placebos, on longitudinal depression outcomes by using multiplicative interaction terms (median BDNF category-x-treatment); stratified results were presented. Analyses of mediation of treatment effects on change in PHQ-9 by a 2-year change in serum BDNF were also pre-specified using %Mediate macro (Lin et al., 1997).
2.5.7. Post-hoc analyses
1) To mitigate potential bias due to the MDD case selection, we reran the longitudinal analyses addressing the relation of baseline serum BDNF and change in PHQ-9 over 2 years and of treatment effects on change in serum BDNF over 2 years in a random stratified sample; 2) The relation of baseline serum BDNF and change in PHQ-8 over 5 years (long-term change) was examined using repeated measures negative binomial regression; 3) We probed modification by baseline serum BDNF for treatment effects on long-term change in PHQ-8 by using multiplicative interaction terms (BDNF quartile category x treatment); stratified results were presented. We conducted mediation analyses by a 2-year change in serum BDNF for treatment effects on long-term change in PHQ-8 (Lin et al., 1997); 4) We addressed whether the treatment effects on 2-year change in serum BDNF differed across age groups (<65 vs. 65+ years), sex (female vs. male), or race/ethnicity (Black vs. non-Black adults); p-interaction was reported from the test of the subgroup-x-treatment-x-follow-up time interaction term in the model; 5) There is a lack of an evidence-based optimal cut-point of serum BDNF for LLD prediction. Thus, we employed a robust bootstrap technique for receiver operating characteristic (ROC) curve analysis to evaluate the performance of serum BDNF for discriminating incident vs. no incident MDD (Thiele and Hirschfeld, 2021); the longitudinal analysis addressing the relation of serum BDNF to incident MDD was rerun using an optimal cut-point of baseline serum BDNF.
Statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, USA), and R. Tests were two-sided; p<0.05 was used for statistical significance. There was no control for multiple hypothesis testing, and no formal adjustment was made to the p-values. Thus, exploratory analyses of modification and moderation effects of serum BDNF, and post-hoc analyses, including those regarding subgroups, should be interpreted with caution.
3. RESULTS
3.1. Baseline characteristics
In the total sample (n=400), baseline serum BDNF was normally distributed [mean (standard deviation): 36.4 (7.6) ng/mL]; the median age (interquartile range) was 65.1 (60.4-69.2) years; 50.5% were females; and 20.8% were racial/ethnic minorities. Table 1 represents descriptive characteristics by baseline depression status in the sample eligible for cross-sectional analyses (n=380). Compared to those without any history of major psychiatric disorders, participants with a history of depression were slightly younger, had higher body mass index; reported lower leisure-time physical activity, had higher percentages of past/current cigarette smoking, ~2-fold higher prevalence of diabetes, and higher median PHQ-9 score. Notably, there was no difference in baseline serum BDNF by depression status.
Table 1.
Participants’ characteristics in a random sample at baseline, stratified by depression case status.
| Baseline characteristics | Sample | Depression casesa | Non-depressed individuals | P-value |
|---|---|---|---|---|
| (n=380) | (n=100) | (n=280) | ||
| Age at CTSC, median (IQR) | 64.9 (60.3 - 69.2) | 63.6 (58.8 - 66.9) | 65.3 (60.7 - 69.5) | 0.01b |
| Sex, n (%) | 0.73 | |||
| Male | 188 (49.5) | 48 (48.0) | 140 (50.0) | |
| Female | 192 (50.5) | 52 (52.0) | 140 (50.0) | |
| Race/ethnicity, n (%) | 0.98 | |||
| Non-Hispanic white | 297 (78.2) | 78 (78.0) | 219 (78.2) | |
| Black | 40 (10.5) | 11 (11.0) | 29 (10.4) | |
| Othersd | 43 (11.3) | 11 (11.0) | 32 (11.4) | |
| Education, n (%) | 0.49 | |||
| ≤college degree | 182 (48.0) | 51 (51.0) | 131 (47.0) | |
| Post-college education | 197 (52.0) | 49 (49.0) | 148 (53.1) | |
| Income, n (%) | 0.18 | |||
| <$50,000 | 77 (23.3) | 25 (28.4) | 52 (21.4) | |
| $50,000+ | 254 (76.7) | 63 (71.6) | 191 (78.6) | |
| BMI, kg/m2, median (IQR) | 27.5 (24.4 – 30.6) | 28.2 (25.1 - 31.6) | 27.3 (24.4 – 29.9) | 0.05b |
| PA,e MET-hours/week, median (IQR) | 21.5 (8.0 - 38.0) | 18.3 (7.2 - 30.6) | 21.7 (8.2 - 40.6) | 0.07b |
| Alcohol use, n (%) | 0.13 | |||
| Rare/never | 69 (19.3) | 25 (26.3) | 44 (16.7) | |
| Monthly | 32 (8.9) | 7 (7.4) | 25 (9.5) | |
| Weekly | 154 (43.0) | 42 (44.2) | 112 (42.6) | |
| Daily | 103 (28.8) | 21 (22.1) | 82 (31.2) | |
| Smoking use, n (%) | 0.03 | |||
| Never | 197 (52.1) | 42 (42.0) | 155 (55.8) | |
| Past | 163 (43.1) | 50 (50.0) | 113 (40.7) | |
| Current | 18 (4.8) | 8 (8.0) | 10 (3.6) | |
| History of hypertension,f n (%) | 177 (47.0) | 49 (49.5) | 128 (46.0) | 0.55 |
| History of diabetes,g n (%) | 44 (11.6) | 18 (18.2) | 26 (9.3) | 0.02 |
| History of high cholesterol, n (%) | 140 (36.8) | 42 (42.0) | 98 (35.0) | 0.21 |
| PHQ-9 score, median (IQR) | 1.0 (0.0 - 2.0) | 1.0 (0.0 - 3.0) | 0.0 (0.0 - 2.0) | 0.001b |
| SSRI users, n (%) | 18 (4.8) | 12 (12.0) | 6 (2.2) | <0.001 |
| Serum BDNF, ng/mL, mean (SD) | 36.3 (7.5) | 37.4 (7.2) | 36.0 (7.5) | 0.11c |
| Randomization to vitamin D3 group, n (%) | 0.95 | |||
| Active | 189 (49.7) | 50 (50.0) | 139 (49.6) | |
| Placebo | 191 (50.3) | 50 (50.0) | 141 (50.4) | |
| Randomization to omega-3s group, n (%) | 0.95 | |||
| Active | 191 (50.3) | 50 (50.0) | 141 (50.4) | |
| Placebo | 189 (49.7) | 50 (50.0) | 139 (49.6) |
Abbreviation: BDNF, brain-derived neurotrophic factor; BMI, body mass index; PA, physical activity; PHQ, patient health questionnaire; CTSC, clinical and translational science center; IQR, interquartile range.
Depression cases include current, past, or recurrent major depressive disorder, current or past major depressive episodes and dysthymia.
Wilcoxon-rank sum test was performed.
Two-sample t-test was performed.
Others race/ethnicity group included Hispanics, Asians and Other (Native Hawaiian or other Pacific Islander, multiple race, unknown race, or unknown ethnicity).
Leisure-time physical activities include walking or hiking; jogging; running; bicycling; aerobic exercise/aerobic dance/exercise machines; lower intensity exercise/yoga/stretching/toning; tennis/squash/racquetball; lap swimming; weight lifting/strength training; other exercise.
Ever diagnosed with high blood pressure or ever use of anti-hypertensive medication
Ever diagnosed with diabetes or current use of anti-diabetic medication.
Table S1 presents descriptive characteristics by quartiles of serum BDNF in the sample eligible for longitudinal analyses (n=288). Females had significantly higher serum BDNF than males (p-trend <0.001). There were no noticeable differences in serum BDNF by other demographic, lifestyle/behavioral, health-related, or depression variables.
3.2. Cross-sectional relations of serum BDNF to depression outcomes
Table 2A shows the mean difference in serum BDNF between participants with a history of depression vs. those without a history of major psychiatric disorders at baseline; there was no significant difference between the groups [adjusted regression coefficient (β) (95% CI): 1.11 (−0.59 to 2.80); p=0.20]. Table 2B shows results for differences in total depressive symptom levels, as indicated by the PHQ-9 score, by quartiles of serum BDNF; participants in the lowest vs. highest quartile of serum BDNF (14.9 – 31.5 ng/mL vs. 40.9 – 69.2 ng/mL) had an 80% higher severity of depressive symptoms after adjusting for age, sex, Black vs. non-Black, history of diabetes and SSRI use (percent differences=80%; 95% CI (12% to 188%); p-value=0.01); overall trend across serum BDNF quartiles was non-monotonic and appeared meaningful (p-trend=0.06).
Table 2A.
Mean difference (95% CI) in serum BDNF (in ng/mL) comparing depression cases (n=100) vs. those without any history of psychiatric disorders (n=280).
| Outcome: depression casesa vs. those without any history of psychiatric disorders |
Mean difference (95% CI) | P-value |
|---|---|---|
| Model 1b | 1.40 (−0.31 to 3.11) | 0.11 |
| Model 2c | 1.11 (−0.59 to 2.80) | 0.20 |
Abbreviation: BDNF, brain-derived neurotrophic factor; CI, confidence interval
Depression cases include current, past, or recurrent major depressive disorder, current or past major depressive episodes, and dysthymia.
Univariate model.
Model adjusted by age, sex, Blacks vs. non-Blacks, history of diabetes, and SSRI use.
Table 2B.
Percent differences in total depression severity (PHQ-9 score) according to quartiles of baseline serum BDNF (n=380).*
| Quartiles of serum BDNF at baseline | |||||
|---|---|---|---|---|---|
| Highest (Q4) (n=95) 40.9 – 69.2 ng/mL |
Third (Q3) (n=95) 35.6 – 40.9 ng/mL |
Second (Q2) (n=95) 31.5 – 35.5 ng/mL |
Lowest (Q1) (n=95) 14.9 – 31.5 ng/mL |
P-value for trend | |
| Percent differences (95% CI) | |||||
| Model la | 0% (Ref) | 53% (−3% to 140%) | −7% (−42% to 49%) | 54% (−2% to 142%) | 0.27 |
| Model 2b | 0% (Ref) | 54% (−3% to 145%) | 9% (−33% to 77%) | 80% (12% to 188%) | 0.06 |
Abbreviation: RR, rate ratio; BDNF, brain-derived neurotrophic factor; PHQ, patient health questionnaire; CI, confidence interval
Results from the negative binomial (NB) regression model show percent differences and 95% CIs in total depression severity on the PHQ-9 by quartiles of serum BDNF at baseline.
Univariate model.
Model adjusted by age, sex, race/ethnicity, history of diabetes, and SSRI use.
3.3. Longitudinal relations of baseline serum BDNF levels to depression outcomes over 2 years
In the sample eligible for longitudinal analysis (n=288), over 2 years, 28 participants developed incident MDD, and 260 participants remained depression-free. The median serum BDNF at baseline was 35.5 ng/mL. Comparing participants with above vs. below the median of baseline serum BDNF levels, there was no statistically significant difference in the likelihood of incident MDD (Table 3A). Furthermore, being in the lowest vs. highest quartiles of baseline serum BDNF was not associated with a mean difference in change in PHQ-9 over 2 years (p-trend=0.82) (Table 3B).
Table 3A.
Odds ratio and 95% CI of DSM-IV incident MDD cases over 2 years by median category of baseline serum BDNF.
| Baseline serum BDNF | Cases (n=28) | Non cases (n=260) | OR (95% CI)a | OR (95% CI)b |
|---|---|---|---|---|
| <35.5 ng/mL (n=144) | 11 | 133 | 1.00 (Ref) | 1.00 (Ref) |
| ≥35.5 ng/mL (n=144) | 17 | 127 | 1.62 (0.73 to 3.59) | 1.81 (0.78 to 4.22) |
Abbreviation: DSM, Diagnostic and Statistical Manual of Mental Disorders; BDNF, brain-derived neurotrophic factor; MDD, major depressive disorder; CI, confidence interval.
Univariate model
Models were adjusted by age at CTSC visit, sex, and treatment agents.
Table 3B.
Difference in change in PHQ-9 score over 2-year follow-up according to quartiles of serum BDNF at baseline (n=288).
| Quartiles of serum BDNF at baseline | |||||
|---|---|---|---|---|---|
| Highest (Q4) (n=72) 40.9 – 69.2 ng/mL |
Third (Q3) (n=72) 35.5 – 40.7 ng/mL |
Second (Q3) (n=72) 31.1 – 35.5 ng/mL |
Lowest (Q1) (n=72) 14.9 – 31.1 ng/mL |
P-value for trend | |
| RR (95% CI)a | |||||
| Model 1b | 1.00 (Ref) | 0.68 (0.43 to 1.08) | 1.03 (0.65 to 1.65) | 0.82 (0.53 to 1.27) | 0.87 |
| Model 2c | 1.00 (Ref) | 0.67 (0.42 to 1.07) | 1.02 (0.63 to 1.63) | 0.81 (0.52 to 1.25) | 0.82 |
Abbreviation: RR, rate ratio; PHQ, patient health questionnaire; BDNF, brain-derived neurotrophic factor; CI, confidence interval.
Analyses were from repeated measures negative binomial regression models, with follow-up time modeled as an indicator. Results show RR and 95% CI, which reflect differences in the change in severity on the PHQ-9 score over 2 years comparing quartiles of serum BDNF at baseline.
Univariate model.
Models were adjusted by age at CTSC visit, sex, and treatment agents.
3.4. Effects of treatment agents on change in serum BDNF levels over 2 years
We did not observe a statistically significant mean difference in 2-year change in serum BDNF levels among those randomized to treatment agents vs. placebos [adjusted mean difference (95% CI)=0.51 (−0.59 to 1.60) for vitamin D3 vs. placebo; −0.62 (−1.71 to 0.48) for omega-3s vs. placebo] (Table 4).
Table 4.
Adjusted mean difference in change in serum BDNF (in ng/mL) over 2 years comparing treatment agents to their matching placebos (n=288).
| Treatment groups | Adjusted mean difference (95% CI)a | P-value |
|---|---|---|
| Vitamin D3 vs. placebo | 0.51 (−0.59 to 1.60) | 0.36 |
| Omega-3s vs. placebo | −0.62 (−1.71 to 0.48) | 0.27 |
Abbreviation: BDNF, brain-derived neurotrophic factor; CI, confidence interval.
Results from the repeated measures response profile model show mean differences in the change in serum BDNF over 2 years comparing treatment agents and placebos, after controlling for age, sex, and concurrent randomization group; all participants contributed to the repeated measure analysis at one and/or both time points.
3.5. Exploratory analyses of modification and mediation effects of serum BDNF
Modification analyses revealed that baseline serum BDNF (i.e., above vs. below median) did not significantly modify the estimates for the effects of treatment agents vs. placebos on risk of MDD or on mean differences in change in PHQ-9 over 2 years (all p-interactions>0.05) (Table 5A and 5B). Because we did not observe any evidence in this sample of effects of vitamin D3 or omega-3s on 2-year change in PHQ-9 or of an association between 2-year change in serum BDNF and 2-year change in PHQ-9 (results are not shown here), we concluded that 2-year change in serum BDNF did not mediate the treatment effects for 2-year change in mood.
Table 5A.
Effect of vitamin D3 or omega-3s, compared to their matching placebos, on the likelihood of DSM-IV incident MDD over 2 years by median category of serum BDNF.
| 1) Vitamin D3 vs. placebo | ||||
|---|---|---|---|---|
| Vitamin D3 group | Placebo group | OR (95% CI)a | P-interaction | |
| Cases/no. of participants |
Cases/no. of participants |
|||
| Baseline serum BDNF | 0.41 | |||
| <35.5 ng/mL (n=144) | 4/76 | 7/68 | 0.48 (0.13 to 1.76) | |
| ≥35.5 ng/mL (n=144) | 7/65 | 10/79 | 1.17 (0.39 to 3.53) | |
| 2) Omega 3s vs. placebo | ||||
| Omega-3s group | Placebo group | OR (95% CI)a | P-interaction | |
| Cases/no. of participants |
Cases/no. of participants |
|||
| Baseline serum BDNF | 0.31 | |||
| <35.5 ng/mL (n=144) | 5/71 | 6/73 | 0.80 (0.23 to 2.80) | |
| ≥35.5 ng/mL (n=144) | 11/75 | 6/69 | 1.93 (0.65 to 5.77) | |
Abbreviation: BDNF, brain-derived neurotrophic factor; MDD, major depressive disorder; DSM, diagnostic and statistical manual of mental disorders; OR, odds ratio; CI, confidence interval.
Model was adjusted by age at CTSC visit, sex, and randomization to omega-3s group.
Model was adjusted by age at CTSC visit, sex, and randomization to vitamin D3 group.
Table 5B.
Effect of vitamin D3 or omega-3s, compared to their matching placebos, on mean difference in change in PHQ-9 score over 2 years by quartiles of serum BDNF at baseline.
| Vitamin D3 vs. placebo | Omega-3s vs. placebo | |||
|---|---|---|---|---|
| Mean difference (95% CI) |
P-interaction | Mean difference (95% CI) |
P-interaction | |
| Quartiles of serum BDNF | 0.89 | 0.42 | ||
| Lowest (Q1) (n=72) | 0.99 (0.53 to 1.85) | 0.91 (0.50 to 1.65) | ||
| Second (Q2) (n=72) | 0.87 (0.42 to 1.79) | 1.37 (0.71 to 2.66) | ||
| Third (Q3) (n=72) | 1.25 (0.66 to 2.36) | 1.48 (0.77 to 2.83) | ||
| Highest (Q4) (n=72) | 0.69 (0.38 to 1.25) | 1.25 (0.69 to 2.26) | ||
Abbreviation: BDNF, brain-derived neurotrophic factor; PHQ, patient health questionnaire; CI, confidence interval Analyses were from repeated measures negative binomial regression models, with follow-up time modeled as an indicator. Results show RR and 95% CI, which reflect differences in the change in serum BDNF over 2 years comparing treatment agents and placebos. Models were controlled for age, sex, and concurrent randomization group. P-interaction is from the test of the BDNF quartile category-x-treatment-x-time interaction term in the model.
3.6. Post-hoc analyses
These results showed: 1) Results of baseline serum BDNF and 2-year change in PHQ-9 and of treatment effects on 2-year change in BDNF were non-significant in a random stratified sample (results are not shown); the estimates were comparable to our primary analyses and may mitigate the possibility of bias due to the MDD case selection; 2) Being in the lowest vs. highest quartiles of baseline serum BDNF was not associated with long-term (5+ years) change in PHQ-8 (Table S2); 3) baseline serum BDNF did not significantly modify the effects of vitamin D3 or omega-3s on average mean difference in change in PHQ-8 score over 5 years of follow-up (Table S3). Further, because we did not observe any evidence of treatment effects on long-term change in mood, or of an association between a 2-year change in serum BDNF and a 5-year change in PHQ-8 (results are not shown), we concluded that a 2-year change in BDNF did not mediate the treatment effects for long-term change in mood; 4) No significant variations by age, sex, and race were found in the treatment effects on change in serum BDNF over 2 years (all p-interactions>0.05; stratified results are not shown here); 6) The optimal cut-off of serum BDNF to discriminate MDD cases from those who remained depression-free over 2 years, as determined by ROC analysis (Thiele and Hirschfeld, 2021), was 36.1 ng/mL (area under curve=0.54; sensitivity=60.7%; specificity=55.0%) (Figure S2); the result of optimal serum BDNF category and incident MDD was similarly null to those results of a median category of serum BDNF and incident MDD.
4. DISCUSSION
This study provides new results regarding the longitudinal associations between serum BDNF and LLD, the effects of two nutrient supplements (vitamin D3 and omega-3s) on serum BDNF, and potential modification and mediation by serum BDNF for effects of these supplements on LLD. There were no significant cross-sectional differences in mean serum BDNF comparing participants with a history of depression and those without any history of psychiatric disorders; however, being in the lowest vs. highest quartile of baseline serum BDNF (14.9 – 31.5 ng/mL vs. 40.9 – 69.2 ng/mL) was significantly associated with worse depressive symptoms. In longitudinal analyses, there were no significant associations between serum BDNF and incident MDD or change in PHQ-9 over 2 years. Compared to their placebos, vitamin D3 or omega-3s had no significant effects on change in serum BDNF, and serum BDNF did not appear to modify or mediate effects of either nutraceutical on LLD in exploratory analyses.
Converging lines of evidence suggest that BDNF plays a pivotal role in the pathophysiology of LLD (Duman et al., 2000; Duman and Monteggia, 2006; Karege et al., 2002). Prior findings regarding lower serum BDNF in depressed compared to non-depressed individuals were compelling; yet, other studies have reported null associations (Ohio et al., 2021; Ziegenhorn et al., 2007), and we did not observe differences in mean serum BDNF comparing participants with depression history vs. those without a history of major psychiatric disorders in our sample. Differences in clinical characteristics and study methods may account for heterogeneity in reported findings in the literature (e.g., age and sex distribution of sample; study setting [community-based vs. tertiary-care clinic]; depression outcome used; matching criteria used for sample selection; psychiatric comorbidity in the sample; concomitant use of antidepressants or other psychotropic drugs) (Bocchio-Chiavetto et al., 2010; Bus et al., 2011; Molendijk et al., 2014; Shi et al., 2020; Trajkovska et al., 2007). Separately, our finding suggesting an inverse cross-sectional association between serum BDNF level and depressive symptom severity level is consistent with some prior studies but not all (Caldieraro et al., 2017; Jevtovic et al., 2011; Molendijk et al., 2011). Explanations for such discrepancies may include differences in the mood scales used, sampling of participants at various stages of disease progression, and influences of potential measured or unmeasured confounders (e.g., seasonality, physical activity).
In this study, we also probed the clinical utility of serum BDNF as a potential predictive biomarker for longitudinal depression outcome, but we did not observe any evidence that baseline serum BDNF predicted MDD risk or change in mood scores in this sample. Although the multivariable odds ratio of incident MDD, compared to those who remained depression-free, was not statistically significant, it was nonetheless almost two-fold higher, raising the possibility of a true relationship. Genetic inheritance and epigenetic mechanisms (e.g., DNA methylation and histone modification) that regulate BDNF expression have been implicated in the development of MDD (Hing et al., 2018). Furthermore, emerging literature suggests that variability in BDNF could be associated with aging-related processing and signaling pathways and may promote changes in cellular and molecular aging makers, and thus, could affect the MDD risk (Miranda et al., 2019; Molinari et al., 2020). Although DNA genotyping and cellular and molecular aging markers were not profiled in our sample, the possibility of increased MDD risk through BDNF polymorphisms or aging-related pathways cannot be excluded, and future studies may address these mechanisms.
Our results regarding the effects of vitamin D3 or omega-3s supplementation on 2-year change in serum BDNF represent a novel contribution to the literature and we also newly probed the question of variation by age, sex, and race/ethnicity in this prospective association; however, all results were null. These null findings in a 2-year RCT of adults contrast with experimental rat models, which suggest that vitamin D3 and omega-3s have neuroprotective effects by activating the BDNF transcription pathways (Gravesteijn et al., 2022; Kiraly et al., 2006; Wu et al., 2008). In the literature, there were 3 prior small-scale and short-term intervention studies in generally healthy older adults (n<100; intervention period:≤6 months) testing the effects of vitamin D3 or omega-3s supplements on alteration in serum BDNF concentrations, and none found significant intervention effects on change in BDNF (Gravesteijn et al., 2022). Separately, our findings also suggest that serum BDNF did not appear to modify or mediate the longitudinal relations of treatment agents with LLD.
Our study had notable strengths. The cohort is well-characterized, features diverse racial and ethnic minority representation, and has well-validated and rigorously adjudicated depression measures at baseline and follow-up. Another advantage was that we performed BDNF assays in serum rather than plasma samples, since serum BDNF may represent a long-term measure of brain BDNF levels (Fujimura et al., 2002). Also, the study contributes new information to the literature by addressing: the association between baseline serum BDNF and shorter-term (2 years) and longer-term (5+ years) changes in mood; the effects of vitamin D3 or omega-3s supplementations on change in serum BDNF over 2 years; exploration of modification and mediation effects of serum BDNF on the treatment effects on short-term and long-term change in mood.
This study also has limitations. First, the incidence rate of MDD over two years was low (~5%); however, this is consistent with the low incidence of MDD in samples of generally healthy and high-functioning community-recruited (i.e., not a hospital or clinic-based) participants (Bot et al., 2019; de Koning et al., 2019). Second, interpretation of our findings is constrained to the use of serum/blood-based BDNF markers and may not generalize to results that may be observed when using BDNF in cerebrospinal fluid (CSF); it is still unclear whether peripheral BDNF concentrations reflect its levels in the central nervous system. Nonetheless, BDNF has been reported to be able to pass through the blood-brain barrier, and modest, positive correlations have been shown between serum and CSF BDNF levels (Klein et al., 2011). Third, study participants were members of a long-term randomized trial cohort and may therefore be healthier than older adults in the general community; while this may affect generalizability, it does not detract from the internal validity of the findings. Fourth, evidence suggests that doses of ≥1.5 g/day omega-3s may be necessary for LLD prevention (Bai et al., 2018); however, due to the fixed-dose design in VITAL, we were unable to test effects of high-dose omega-3s or of different balances of EPA vs. DHA on change in serum BDNF levels.
In conclusion, we observed a significant cross-sectional association between serum BDNF with severity of depressive symptoms, but not with depression case status. In longitudinal analyses, there were no significant associations between serum BDNF with incident MDD or change in mood over 2 years. Furthermore, daily supplementation of vitamin D3 and/or omega-3s did not alter serum BDNF over 2 years, and serum BDNF did not appear to modify or mediate longitudinal associations between vitamin D3 or omega-3 supplements and LLD.
Supplementary Material
There is an urgent need to identify biomarkers for major depressive disorder.
Low serum BDNF has promise as a biomarker for major depressive disorder.
Low serum BDNF was associated with worse depressive symptoms at baseline.
Serum BDNF did not predict the new onset of major depressive disorder.
Vitamin D and omega-3s did not alter BDNF levels over 2 years.
ACKNOWLEDGMENTS
We acknowledge the invaluable contributions and dedication of the 25,871 participants in VITAL and the entire staff of the VITAL study.
VITAL-DEP is supported by R01 MH091448 and R56 MH091448 from the National Institute of Mental Health (NIMH). VITAL is supported by grants R01 AT011729, U01 CA138962, and R01 CA138962, which include support from the National Cancer Institute; National Heart, Lung and Blood Institute (NHLBI); Office of Dietary Supplements; National Institute of Neurological Disorders and Stroke; and the National Center for Complementary and Integrative Health of the National Institutes of Health (NIH). The VITAL ancillary studies and CTSC (Clinical and Translational Science Center) component are supported by grants DK088078 and R01 DK088762 from the National Institute of Diabetes and Digestive and Kidney Diseases; R01 HL101932 and R01 HL102122 from NHLBI; R01 AG036755 from the National Institute on Aging (NIA); R01 AR059086 and R01 AR060574 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases; and R01 MH091448 from NIMH. This work was conducted with support from the Harvard Catalyst CTSC (UL1TR001102 from the National Center for Advancing Translational Sciences). Dr. Reynolds’ participation also received support from P30 MH090333 from NIMH, and the University of Pittsburgh Medical Center Endowment in Geriatric Psychiatry. Pharmavite LLC of Northridge, California (vitamin D) and Pronova BioPharma/BASF of Norway (Omacor® fish oil) donated the study agents, matching placebos, and packaging in the form of calendar packs. Dr. Mora was supported by research grants from the National Institute of Diabetes and Digestive and Kidney Diseases (DK112940) and NHLBI (R01HL134811 and K24 HL136852, R01HL134168, and 1R01HL143227). VITAL-DEP has been approved by the Institutional Review Board of Partners Healthcare/Brigham and Women’s Hospital, and the VITAL study agents have received Investigational New Drug Approval from the U.S. Food and Drug Administration. Voting members of the Data and Safety Monitoring Board for VITAL and ancillary studies, including VITAL-DEP, included Lawrence S. Cohen, MD; Theodore Colton, ScD; Mark A. Espeland, PhD; Craig Henderson, MD; Alice H. Lichtenstein, ScD; Rebecca A. Silliman, MD, PhD; and Nanette Wenger, MD (chair). Ex-officio members include Josephine Boyington, PhD, MPH; Rebecca Costello, PhD; Cindy Davis, PhD; Peter Greenwald, MD; and Wendy Weber, PhD. VITAL and VITAL-DEP are registered at clinicaltrials.gov (VITAL: NCT01169259; VITAL-DEP: NCT01696435). The VITAL website is www.vitalstudy.org.
Role of the Funder/Sponsor:
The NIH, Harvard Catalyst, US FDA, Pharmavite LLC, and Pronova BioPharma/BASF had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
CONFLICT OF INTEREST
Dr. Vyas has received research support from Nestle-Purina Petcare Company.
Dr. Mischoulon has received research support from Nordic Naturals and Heckel Medizintechnik GmbH. He has received honoraria for speaking from the Massachusetts General Hospital Psychiatry Academy, Peerpoint Medical Education Institute, LLC, and Harvard blog. He also works with the MGH Clinical Trials Network and Institute (CTNI), which has received research funding from multiple pharmaceutical companies and NIMH.
Dr. Chang receives royalties from Up-to-Date.
Dr. Buring's spouse was on the Scientific Advisory Board of Pharmavite LLC during the trial.
Dr. Mora has served as consultant to Pfizer and Quest Diagnostics.
Dr. Manson has received research support from Mars Edge.
Dr. Reynolds receives payment from the American Association of Geriatric Psychiatry as Editor-in-Chief of the American Journal of Geriatric Psychiatry, royalty income for intellectual property as co-inventor of the Pittsburgh Sleep Quality Index and, in the past, a one-time honorarium from Merck for consultation on care pathways for insomnia. Dr. Reynolds also receives royalty income from Oxford University Press and from Up-to-Date.
Dr. Okereke receives royalties from Springer Publishing for a book on late-life depression prevention.
No other authors have disclosures to report.
REFERENCES
- Akaike H. Information theory as an extension of the maximum likelihood principle. Petrov BV, Csaki BF (eds). Second International Symposium on Information Theory. Academiai Kiado; Budapest: 1973. pp. 267–281. [Google Scholar]
- Bai ZG, Bo A, Wu SJ, Gai QY, Chi I, 2018. Omega-3 polyunsaturated fatty acids and reduction of depressive symptoms in older adults: A systematic review and meta-analysis. J Affect Disord. 241, pp. 241–248. [DOI] [PubMed] [Google Scholar]
- Bathina S, Das UN, 2015. Brain-derived neurotrophic factor and its clinical implications. Arch Med Sci. 11 (6), pp. 1164–1178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Björkholm C, Monteggia LM, 2016. BDNF - a key transducer of antidepressant effects. Neuropharmacology. 102, pp. 72–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bocchio-Chiavetto L, Bagnardi V, Zanardini R, Molteni R, Nielsen MG, Placentino A, Giovannini C, Rillosi L, Ventriglia M, Riva MA, Gennarelli M, 2010. Serum and plasma BDNF levels in major depression: a replication study and meta-analyses. World J Biol Psychiatry. 11 (6), pp. 763–773. [DOI] [PubMed] [Google Scholar]
- Bot M, Brouwer IA, Roca M, Kohls E, Penninx B, Watkins E, van Grootheest G, Cabout M, Hegerl U, Gili M, Owens M, Visser M, 2019. Effect of Multinutrient Supplementation and Food-Related Behavioral Activation Therapy on Prevention of Major Depressive Disorder Among Overweight or Obese Adults With Subsyndromal Depressive Symptoms: The MooDFOOD Randomized Clinical Trial. Jama. 321 (9), pp. 858–868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Briones TL, Darwish H, 2012. Vitamin D mitigates age-related cognitive decline through the modulation of pro-inflammatory state and decrease in amyloid burden. J Neuroinflammation. 9, pp.244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bursac Z, Gauss CH, Williams DK, Hosmer DW, 2008. Purposeful selection of variables in logistic regression. Source Code Biol Med. 3, pp. 17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bus BA, Molendijk ML, Penninx BJ, Buitelaar JK, Kenis G, Prickaerts J, Elzinga BM, Voshaar RC, 2011. Determinants of serum brain-derived neurotrophic factor. Psychoneuroendocrinology. 36 (2), pp. 228–239. [DOI] [PubMed] [Google Scholar]
- Caldieraro MA, Vares EA, Souza LH, Spanemberg L, Guerra TA, Wollenhaupt-Aguiar B, Ferrari P, Nierenberg AA, Fleck MP, 2017. Illness severity and biomarkers in depression: Using a unidimensional rating scale to examine BDNF. Compr Psychiatry. 75, pp. 46–52. [DOI] [PubMed] [Google Scholar]
- Crombie KM, Sartin-Tarm A, Sellnow K, Ahrenholtz R, Lee S, Matalamaki M, Almassi NE, Hillard CJ, Koltyn KF, Adams TG, Cisler JM, 2021. Exercise-induced increases in Anandamide and BDNF during extinction consolidation contribute to reduced threat following reinstatement: Preliminary evidence from a randomized controlled trial. Psychoneuroendocrinology. 132, pp. 105355. [DOI] [PMC free article] [PubMed] [Google Scholar]
- de Koning EJ, Lips P, Penninx B, Elders PJM, Heijboer AC, den Heijer M, Bet PM, van Marwijk HWJ, van Schoor NM, 2019. Vitamin D supplementation for the prevention of depression and poor physical function in older persons: the D-Vitaal study, a randomized clinical trial. Am J Clin Nutr. 110 (5), pp. 1119–1130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Duman RS, Malberg J, Nakagawa S, D'Sa C, 2000. Neuronal plasticity and survival in mood disorders. Biol Psychiatry. 48 (8), pp. 732–739. [DOI] [PubMed] [Google Scholar]
- Duman RS, Monteggia LM, 2006. A neurotrophic model for stress-related mood disorders. Biol Psychiatry. 59 (12), pp. 1116–1127. [DOI] [PubMed] [Google Scholar]
- Edinoff AN, Akuly HA, Hanna TA, Ochoa CO, Patti SJ, Ghaffar YA, Kaye AD, Viswanath O, Urits I, Boyer AG, Cornett EM, Kaye AM, 2021. Selective Serotonin Reuptake Inhibitors and Adverse Effects: A Narrative Review. Neurol Int. 13 (3), pp. 387–401. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fujimura H, Altar CA, Chen R, Nakamura T, Nakahashi T, Kambayashi J, Sun B, Tandon NN, 2002. Brain-derived neurotrophic factor is stored in human platelets and released by agonist stimulation. Thromb Haemost. 87 (4), pp. 728–734. [PubMed] [Google Scholar]
- Gravesteijn E, Mensink RP, Plat J, 2022. Effects of nutritional interventions on BDNF concentrations in humans: a systematic review. Nutr Neurosci. 25 (7), pp. 1425–1436. [DOI] [PubMed] [Google Scholar]
- Hing B, Sathyaputri L, Potash JB, 2018. A comprehensive review of genetic and epigenetic mechanisms that regulate BDNF expression and function with relevance to major depressive disorder. Am J Med Genet B Neuropsychiatr Genet. 177 (2), pp. 143–167. [DOI] [PubMed] [Google Scholar]
- Jevtović S, Karlović D, Mihaljević-Peleš A, Šerić V, Vrkić N, Jakšić N, 2011. Serum Brain-derived neurotrophic factor (BDNF): the severity and symptomatic dimensions of depression. Psychiatr Danub. 23 (4), pp. 363–369. [PubMed] [Google Scholar]
- Karege F, Schwald M, Cisse M, 2002. Postnatal developmental profde of brain-derived neurotrophic factor in rat brain and platelets. Neurosci Lett. 328 (3), pp. 261–264. [DOI] [PubMed] [Google Scholar]
- Khairy EY, Attia MM, 2021. Protective effects of vitamin D on neurophysiologic alterations in brain aging: role of brain-derived neurotrophic factor (BDNF). Nutr Neurosci. 24 (8), pp. 650–659. [DOI] [PubMed] [Google Scholar]
- Kiraly SJ, Kiraly MA, Hawe RD, Makhani N, 2006. Vitamin D as a neuroactive substance: review. ScientificWorldJoumal. 6, pp. 125–139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Klein AB, Williamson R, Santini MA, Clemmensen C, Ettrup A, Rios M, Knudsen GM, Aznar S, 2011. Blood BDNF concentrations reflect brain-tissue BDNF levels across species. Int J Neuropsychopharmacol. 14 (3), pp. 347–353. [DOI] [PubMed] [Google Scholar]
- Knöchel C, Voss M, Grüter F, Alves GS, Matura S, Sepanski B, Stäblein M, Wenzler S, Prvulovic D, Carvalho AF, Oertel-Knöchel V, 2015. Omega 3 Fatty Acids: Novel Neurotherapeutic Targets for Cognitive Dysfunction in Mood Disorders and Schizophrenia? Curr Neuropharmacol. 13 (5), pp. 663–680. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kroenke K, Spitzer RF, Williams JB, 2001. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med. 16 (9), pp. 606–613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kroenke K, Strine TW, Spitzer RF, Williams JB, Berry JT, Mokdad AH, 2009. The PHQ-8 as a measure of current depression in the general population. J Affect Disord. 114 (1-3), pp. 163–173. [DOI] [PubMed] [Google Scholar]
- Liao Y, Xie B, Zhang H, He Q, Guo L, Subramanieapillai M, Fan B, Lu C, McIntyre RS, 2019. Efficacy of omega-3 PUFAs in depression: A meta-analysis. Transl Psychiatry. 9 (1), pp. 190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lin DY, Fleming TR, De Gruttola V, 1997. Estimating the proportion of treatment effect explained by a surrogate marker. Stat Med. 16 (13), pp. 1515–1527. [DOI] [PubMed] [Google Scholar]
- Manson JE, Bassuk SS, Lee IM, Cook NR, Albert MA, Gordon D, Zaharris E, Macfadyen JG, Danielson E, Lin J, Zhang SM, Buring JE, 2012. The VITamin D and OmegA-3 TriaL (VITAL): rationale and design of a large randomized controlled trial of vitamin D and marine omega-3 fatty acid supplements for the primary prevention of cancer and cardiovascular disease. Contemp Clin Trials. 33 (1), pp. 159–171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Miranda M, Morici JF, Zanoni MB, Bekinschtein P, 2019. Brain-Derived Neurotrophic Factor: A Key Molecule for Memory in the Healthy and the Pathological Brain. Front Cell Neurosci. 13, pp. 363. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Molendijk ML, Bus BA, Spinhoven P, Penninx BW, Kenis G, Prickaerts J, Voshaar RC, Elzinga BM, 2011. Serum levels of brain-derived neurotrophic factor in major depressive disorder: state-trait issues, clinical features and pharmacological treatment. Mol Psychiatry. 16 (11), pp. 1088–1095. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Molendijk ML, Spinhoven P, Polak M, Bus BA, Penninx BW, Elzinga BM, 2014. Serum BDNF concentrations as peripheral manifestations of depression: evidence from a systematic review and meta-analyses on 179 associations (N=9484). Mol Psychiatry. 19 (7), pp. 791–800. [DOI] [PubMed] [Google Scholar]
- Molinari C, Morsanuto V, Ruga S, Notte F, Farghali M, Galla R, Uberti F, 2020. The Role of BDNF on Aging-Modulation Markers. Brain Sci. 10 (5). [DOI] [PMC free article] [PubMed] [Google Scholar]
- Nagatsu T, Sawada M, 2005. Inflammatory process in Parkinson's disease: role for cytokines. Curr Pharm Des. 11 (8), pp. 999–1016. [DOI] [PubMed] [Google Scholar]
- Ohto A, Mizoguchi Y, Imamura Y, Kojima N, Yamada S, Monji A, 2021. No association of both serum pro-brain-derived neurotrophic factor (proBDNF) and BDNF concentrations with depressive state in community-dwelling elderly people. Psychogeriatrics. 21 (4), pp. 503–513. [DOI] [PubMed] [Google Scholar]
- Okereke OI, Reynolds CF 3rd, Mischoulon D, Chang G, Cook NR, Copeland T, Friedenberg G, Buring JE, Manson JE, 2018. The VITamin D and OmegA-3 TriaL-Depression Endpoint Prevention (VITAL-DEP): Rationale and design of a large-scale ancillary study evaluating vitamin D and marine omega-3 fatty acid supplements for prevention of late-life depression. Contemp Clin Trials. 68, pp. 133–145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Papakostas GI, Fava M, 2008. Predictors, moderators, and mediators (correlates) of treatment outcome in major depressive disorder. Dialogues Clin Neurosci. 10 (4), pp. 439–451. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, Hergueta T, Baker R, Dunbar GC, 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. J Clin Psychiatry. 59 Suppl 20, pp. 22–33;quiz 34-57. [PubMed] [Google Scholar]
- Shi Y, Luan D, Song R, Zhang Z, 2020. Value of peripheral neurotrophin levels for the diagnosis of depression and response to treatment: A systematic review and meta-analysis. Eur Neuropsychopharmacol. 41, pp. 40–51. [DOI] [PubMed] [Google Scholar]
- Strawbridge R, Young AH, Cleare AJ, 2017. Biomarkers for depression: recent insights, current challenges and future prospects. Neuropsychiatr Dis Treat. 13, pp. 1245–1262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Thiele C, Hirschfeld G. cutpointr: Improved Estimation and Validation of Optimal Cutpoints in R. J. Stat. Soft 2021. Jun. 30;98(11):1–27. Available from: https://www.jstatsoft.org/index.php/jss/article/view/v098i11. [Google Scholar]
- Trajkovska V, Marcussen AB, Vinberg M, Hartvig P, Aznar S, Knudsen GM, 2007. Measurements of brain-derived neurotrophic factor: methodological aspects and demographical data. Brain Res Bull. 73 (1-3), pp. 143–149. [DOI] [PubMed] [Google Scholar]
- Vellekkatt F, Menon V, 2019. Efficacy of vitamin D supplementation in major depression: A meta-analysis of randomized controlled trials. J Postgrad Med. 65 (2), pp. 74–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Vuong Q. Likelihood ratio tests for model selection and non-nested hypothesis. Econometrica. 1989;57:307–334. [Google Scholar]
- Wu A, Ying Z, Gomez-Pinilla F, 2008. Docosahexaenoic acid dietary supplementation enhances the effects of exercise on synaptic plasticity and cognition. Neuroscience. 155 (3), pp. 751–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhou C, Zhong J, Zou B, Fang L, Chen J, Deng X, Zhang L, Zhao X, Qu Z, Lei Y, Lei T, 2017. Meta-analyses of comparative efficacy of antidepressant medications on peripheral BDNF concentration in patients with depression. PLoS One. 12 (2), pp. e0172270. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ziegenhorn AA, Schulte-Herbrüggen O, Danker-Hopfe H, Malbranc M, Hartung HD, Anders D, Lang UE, Steinhagen-Thiessen E, Schaub RT, Hellweg R, 2007. Serum neurotrophins--a study on the time course and influencing factors in a large old age sample. Neurobiol Aging. 28 (9), pp. 1436–1445. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.