Table 1 Baseline Characteristics of Included Studies
Study ID | Patients | Controls | Main significant findings | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
Author, Year | Country | Study design | BDNF measurement method | Source (serum, plasma) | No. (Type of ED) | Diagnostic Criteria | BDNF levels (mean ± SD), ng/ml | No | BDNF levels (mean ± SD), ng/ml | |
Nakazato et al., 2002 | Japan | Cross-sectional | BDNF Emax Immunoassay System kit (Promega, Madison, WI) | Serum | 12 (AN-R, AN-BP) | DSM-IV | 24.9 ± 6.75 | 21 | 61.4 ± 19.5 | Lower BDNF levels in AN and BN patients compared to controls Lower BDNF levels in AN patients compared to BN patients Positive correlation between BDNF levels and BMI in all subjects |
Nakazato et al., 2002 | Japan | Cross-sectional | BDNF Emax Immunoassay System kit (Promega, Madison, WI) | Serum | 18 (BN) | DSM-IV | 38.4 ± 15.3 | 21 | 61.4 ± 19.5 | |
Nakazato et al., 2006 | Japan | Longitudinal | BDNF Emax Immunoassay System kit (Promega, Madison, WI, USA) | Serum | 13 (AN-R, AN-BP) | DSM-IV | 14.5 ± 4.4 | 17 | 14.5 ± 4.4 | Lower BDNF levels in AN patients compared to controls No difference in BDNF levels in AN patients before and after partial weight recovery Positive correlation between BDNF levels and ED symptom scores (EDI-2) in all subjects Positive correlation between BDNF levels and BMI in all subjects |
Nakazato et al., 2009 | Japan | Cross-sectional | Emax Immunoassay System kit (Promega, Madison, WI, USA) | Serum | 29 (AN-R, AN-BP) | DSM-IV | 11.7 ± 4.9 | 28 | 15.1 ± 5.5 | Lower BDNF levels in AN patients compared to controls Lower BDNF levels in AN patients compared to patients who recovered from AN Higher rate of set-shifting errors (WCST) in AN patients No correlation between BDNF levels and WCST performance |
Yamada et al., 2012 | Japan | Cross-sectional and longitudinal | ELISA kit (Quantikine, R & D Systems, Minneapolis, MN, USA) | Plasma | 16 (BN) | DSM-IV | 1.89 ± 1.67 | 10 | 6.57 ± 6.09 | Lower BDNF levels in BN patients compared to controls Increased BDNF levels following inpatient treatment, suggesting that lower BDNF in BN is associated with abnormal eating behaviors, especially binge eating |
Dmitrzak-Weglarz et al., 2013 | Poland | Cross-sectional | DuoSet ELISA Development Kits (R&D Systems) – | Serum | 46 (AN-R) | ICD-10 and DSM-IV | 23.72955 ± 8.245963 | 45 | 22.22206 ± 5.938468 | No difference in BDNF levels in AN patients compared with controls No correlation between BDNF levels and BMI or severity of depression symptoms (BDI) in all subjects Correlations between BDNF levels and patient personality dimensions (TCI) |
Dmitrzak-Weglarz et al., 2013 | Poland | Cross-sectional | DuoSet ELISA Development Kits (R&D Systems) – | Serum | 14 (AN-BP) | ICD-10 and DSM-IV | 23.32821 ± 8.078747 | 45 | 22.22206 ± 5.938468 | |
Zwipp et al., 2014 | Germany | Cross-sectional and longitudinal | Enzyme-Linked Immunosorbent Assay kits (ELISA; 181 Promega Inc., Madison, WI, USA) | Serum | 55 (AN) | DSM-IV | 6.6165 ± 3.4158 | 52 | 6.7008 ± 2.7814 | Higher BDNF levels in patients who recovered from AN compared to acutely underweight AN patients Increased BDNF levels with short-term weight gain in acutely underweight AN patients Inverse association of BDNF with psychomotor speed (TMT) in acutely underweight AN patients but not in controls Acutely underweight AN patients with higher BDNF also had lower lifetime BMI, indicating that serum BDNF levels in patients with AN vary with the stage of illness Changes in BDNF levels may have different context-dependent effects, including the modulation of cognitive functioning in acutely underweight patients; as BDNF has pleiotropic functions |
Eddy et al., 2015 | USA | Cross-sectional | immunoassay (ELISA, R&D Systems, Inc) | Serum | 50 (AN-R) | DSM-V | 13.6 ± 0.9 | 22 | 14.6 ± 0.8 | Higher BDNF levels in AN-BP than AN-R No difference in BDNF levels in AN-BP compared to AN-R after controlling for BMI No difference in BDNF levels in AN patients compared to controls No correlation between BDNF levels and BMI in all subjects Positive association between BDNF levels and frequency of purging in all subjects |
Eddy et al., 2015 | USA | Cross-sectional | immunoassay (ELISA, R&D Systems, Inc) | Serum | 25 (AN-BP) | DSM-V | 17.2 ± 1.5 | 22 | 14.6 ± 1.4 | |
Homan et al., 2015 | Switzerland | Randomized, double-blind, placebo-controlled, crossover study | BDNF Emax Immunoassay Kit (Promega, Switzerland) | Plasma | 20 (Remitted BN) | DSM-IV | 3.058 ± 1.5066 | 27 | 2.383 ± 1.2814 | Positive correlation between AMPT–induced differences in BDNF levels with the AMPT–induced differences in reward learning across sample Higher BDNF levels in patients with remitted BN compared to controls across conditions Higher BDNF levels in the morning before compared with after a standardized breakfast across groups and conditions |
Rybakowski et al., 2017 | Poland | Longitudinal | n/a | Serum | 76 (AN) | n/a | 28.94 ± 7.19 | 30 | 34.66 ± 7.4 | Lower BDNF levels in acute AN Normalization of BDNF levels after weight recovery in AN Negative correlation between BDNF levels and body weight in acute AN but not AN after partial weight recovery No correlation between BDNF levels and psychopathological symptoms (HDRS, BDI and YBOCS) in either acute AN or AN after partial weight recovery No evidence found to support a role of serum BDNF levels in the modulation of depressive and obsessive–compulsive symptoms of AN |
Matsumoto et al., 2017 | Japan | Cross-sectional | Human BDNF ELISA kits (Aviscera Bioscience, Santa Clara, CA, USA) | Serum | 19 (AN-R, AN-BP) | DSM-IV | 13.96 ± 6.02 | 22 | 16.58 ± 3.5 | No difference in proBDNF and mBDNF levels among AN patients, BN patients, and controls Correlation between proBDNF and MMP-9 levels in both ED patients and controls Positive correlation between mBDNF levels and IGT performance in BN patients but not in AN patients |
Matsumoto et al., 2017 | Japan | Cross-sectional | Human BDNF ELISA kits (Aviscera Bioscience, Santa Clara, CA, USA) | Serum | 28 (BN) | DSM-IV | 14.99 ± 4.9 | 22 | 12.57 ± 12.54 | |
Mancuso et al., 2020 | USA | Cross-sectional | ELISA (EMD Millipore: Billerica, MA) | Serum | 36 (AN-R, AN-BP) | K-SADS PL and EDE | 2.7313 ± 0.4153 | 32 | 3.7928 ± 0.4575 | Lower fasting BDNF levels in AN patients compared with controls Lower BDNF AUC after breakfast in AN patients compared with controls Positive association between BDNF AUC and kilocalories consumed during CTT in AN patients (particularly AN-R) |
Tyszkiewicz-Nwafor et al., 2020 | Poland | Longitudinal | BDNF DuoSet (cat. No DY 248) and TrkB DuoSet (cat. No DY 397–5) ELISA Development Kit (R&D System, Minneapolis, MN, USA) | Serum | 42 (AN) | ICD-10, DSM-IV and DSM-V | 28.66 ± 6.7 | 30 | 34.66 ± 7.4 | Increased BDNF levels in AN patients after partial weight recovery compared with before, but lower levels compared with controls at two-time points Negative correlation between BDNF levels and the severity of ED symptoms (EAT-26) No correlation between BDNF levels and depressive and obsessive–compulsive symptoms (HDRS, BDI, YBOCS) for either malnourished patients or partially weight recovered AN patients |
Borsdorf et al., 2021 | Germany | Longitudinal | Quantikine ELISA (R&D Systems Inc.) | Serum | 149 (AN) | DSM-IV | 17.36 ± 6.57 | 79 | 14.08 ± 0.75 | Lower BDNF levels in AN patients at admission compared with controls Continuous increase in BDNF levels, reaching supranormal levels at 2.5-year follow-up Inverse association between BDNF levels with ED psychopathology (BDI-II, SCAS, MROAS, SIAB-EX) at discharge Positive association between BDNF levels with previous weight gain at 1-year follow-up |
Keeler et al., 2022 | UK | Cross-sectional | U-PLEX Human BDNF assay (Meso Scale Discovery, Maryland, USA) | Serum | 56 (AN-R, AN-BP) | DSM-V | 5.354 ± 2.616 | 51 | 8.323 ± 2.595 | Lower BDNF levels in AN patients compared to controls and patients recovering from AN Negative association of BDNF with depression and ED psychopathology in the whole sample, but not in AN patients BDNF serum concentrations may be a state marker of AN, but do not reflect symptom severity In acute AN, BDNF levels seem to be linked to TNF-α signaling |