Exercise-induced increases in Anandamide and BDNF during extinction consolidation contribute to reduced threat following reinstatement: Preliminary evidence from a randomized controlled trial

Abstract

Introduction: We recently demonstrated that moderate-intensity aerobic exercise delivered during the consolidation of fear extinction learning reduced threat expectancy during a test of extinction recall among women with posttraumatic stress disorder (PTSD). These findings suggest that exercise may be a potential candidate for improving the efficacy of exposure-based therapies, which are hypothesized to work via the mechanisms of fear extinction learning. The purpose of this secondary analysis was to examine whether exercise-induced increases in circulating concentrations of candidate biomarkers: endocannabinoids (anandamide [AEA]; 2-arachidonoylglycerol [2-AG], brain-derived neurotrophic factor (BDNF), and homovanillic acid (HVA), mediate the effects of exercise on extinction recall.

Methods: Participants (N = 35) completed a 3-day fear acquisition (day 1), extinction (day 2), and extinction recall (day 3) protocol, in which participants were randomly assigned to complete either moderate-intensity aerobic exercise (EX) or a light-intensity control (CON) condition following extinction training (day 2). Blood was obtained prior to and following EX or CON. Threat expectancy ratings during tests of extinction recall (i.e., initial fear recall and fear recall following reinstatement) were obtained 24 h following EX or CON. Mediation was tested using linear-mixed effects models and bootstrapping of the indirect effect.

Results: Circulating concentrations of AEA and BDNF (but not 2-AG and HVA) were found to mediate the relationship between moderate-intensity aerobic exercise and reduced threat expectancy ratings following reinstatement (AEA 95% CI: -0.623 to -0.005; BDNF 95% CI: -0.941 to -0.005).

Conclusions: Exercise-induced increases in peripheral AEA and BDNF appear to play a role in enhancing consolidation of fear extinction learning, thereby leading to reduced threat expectancies following reinstatement among women with PTSD. Future mechanistic research examining these and other biomarkers (e.g., brain-based biomarkers) is warranted.

Keywords: Aerobic exercise; Brain-derived neurotrophic factor; Endocannabinoids; Exposure-therapy; Fear extinction; Posttraumatic stress disorder.

Figure 1.

Graphical overview of the study design. On Day 1, participants completed a fear acquisition task in which the CS+ was paired with a shock (US) with 50% probability. On Day 2 (24hrs later), participants completed a fear extinction task in a distinct context (i.e., different background color) during which the CS+ was no longer paired with US. Immediately after completing the task, half of the participants were randomly assigned to complete 30-min of moderate-intensity aerobic exercise (EX) on a treadmill or 30-min of very-light intensity walking (CON) on a treadmill. Blood draws (to assess circulating concentrations of AEA, 2-AG, BDNF, and HVA) occurred immediately prior to and following the experimental manipulation on day 2. Participants returned on Day 3 (24 hrs later) for a fear extinction recall test alternating between acquisition and extinction contexts (i.e., initial fear recall). A single US was then presented (i.e., reinstatement procedure), followed by another recall test alternating between acquisition and extinction contexts. Threat expectancy ratings during extinction recall were provided once before the initial fear recall test, and immediately after our reinstatement procedure (i.e., prior to the second extinction recall test).

Figure 2.

Circulating concentrations of AEA (A), 2-AG (B), BDNF (C), and HVA (D) prior to and following the light-intensity control condition and moderate-intensity aerobic exercise. In addition to estimating regression coefficients for path a, exploratory post hoc paired samples t-tests were also conducted to further examine circulating concentrations of AEA, 2-AG, BDNF, and HVA prior to and following light-intensity (CON) or moderate-intensity aerobic exercise (EX). There was a significant increase (denoted *) in circulating concentrations of AEA following EX (t(13)=4.21, p = .001, d = 0.58), but not CON (t(13)=−0.66, p = .523, d = −0.16; see figure 2a); along with a significant increase in circulating concentrations of BDNF following EX (t(12)=2.39, p = .034, d = 0.92), but not CON (t(11)=−0.71, p = .494, d = −0.18; see figure 2c); and a significant increase in circulating concentrations of HVA following EX (t(12)=3.71, p = .003, d = 0.94), but not CON (t(11)=1.39, p = .191, d = 0.33; see figure 2d). There was not a significant increase in circulating concentrations of 2-AG following EX (t(13)=−.157, p =.878, d = −0.01) or CON (t(13)=0.922, p=.373, d = 0.20; see figure 2b).

Figure 3.

Mediation Models for AEA and 2-AG. Mediation model for AEA (A) along with corresponding histogram of bootstrapped indirect effects and 95% CI (B) indicated that circulating concentrations of AEA mediated the relationship between aerobic exercise and reduced threat expectancy ratings following reinstatement. Mediation model for 2-AG (C), along with corresponding histogram of bootstrapped indirect effects and 95% CI (D) indicated that circulating concentrations of 2-AG did not mediate the relationship between aerobic exercise and reduced threat expectancy ratings following reinstatement. Values listed are standardized regression coefficients for path a (effect of independent variable on biomarker variable), path b (effect of biomarker variable on dependent variable), the indirect effect (product of regression coefficients for predictor variables in path a and b), path c (effect of independent variable on dependent variable without biomarker variable included in model), and path c’ (difference between direct and indirect effect). 95% CI for the indirect effect (used to determine if mediation occurred; denoted * for 95% CIs not containing zero) was obtained from bootstrap procedure involving resampling with replacement and 10,000 iterations.

Figure 4.

Mediation Models for BDNF and HVA. Mediation model for BDNF (A) along with corresponding histogram of bootstrapped indirect effects and 95% CI (B) indicated that circulating concentrations of BDNF mediated the relationship between aerobic exercise and reduced threat expectancy ratings following reinstatement. Mediation model for HVA (C), along with corresponding histogram of bootstrapped indirect effects and 95% CI (D) indicated that circulating concentrations of HVA did not mediate the relationship between aerobic exercise and reduced threat expectancy ratings following reinstatement. Values listed are standardized regression coefficients for path a (effect of independent variable on biomarker variable), path b (effect of biomarker variable on dependent variable), the indirect effect (product of regression coefficients for predictor variables in path a and b), path c (effect of independent variable on dependent variable without biomarker variable included in model), and path c’ (difference between direct and indirect effect). 95% CI for the indirect effect (used to determine if mediation occurred; denoted * for 95% CIs not containing zero) was obtained from bootstrap procedure involving resampling with replacement and 10,000 iterations.