Brain activation and subjective anxiety during an anticipatory anxiety task is related to clinical outcome during prazosin treatment for alcohol use disorder

Abstract

Background: Higher levels of anxiety, negative affect, and impaired emotion regulation are associated with alcohol use disorder (AUD) and contribute to relapse and worse treatment outcomes. Prazosin, while typically used to treat post-traumatic stress disorder (PTSD) and other anxiety disorders, has shown promise for treating AUD. In order to better understand these underlying neural processes in individuals with AUD, our aims in this study were to measure brain activation during an anticipatory anxiety task before treatment to determine whether observed patterns supported previous work. We then aimed to measure the effects of prazosin on patients with AUD and explore whether greater baseline anticipatory anxiety (as measured by subjective and neural measures) predicts better treatment outcomes.

Methods: Thirty-four individuals seeking treatment for AUD participated in a six-week placebo-controlled study of prazosin and underwent an anticipatory anxiety task during fMRI scans at baseline and three weeks. Alcohol use over six weeks was measured.

Results: Greater levels of subjective anxiety and deactivation in posterior cingulate cortex (PCC) and ventromedial prefrontal cortex (vmPFC) were observed during high-threat stimuli compared to low-threat stimuli. Compared to placebo, prazosin reduced subjective anxiety to high-threat stimuli but there were no observed significant effects of prazosin on brain activation during the task. However, AUD patients with greater vmPFC deactivation during high threat relative to low threat and patients with low baseline anticipatory anxiety during the task had worse clinical outcomes on prazosin.

Conclusions: Deactivation in PCC and vmPFC to high-threat stimuli replicated previous work and shows promise for further study as a marker for AUD. Although prazosin did not affect brain activation in the regions of interest during the anticipatory anxiety task, subjective levels of anxiety and brain activation in vmPFC predicted treatment outcomes in individuals with AUD undergoing treatment with prazosin, highlighting individuals more likely to benefit from prazosin than others.

Keywords: Alcohol use disorder; Anticipatory anxiety task; Prazosin; Stress; Treatment; fMRI.

Fig. 1

This figure is a schemata for the anticipatory anxiety task, and is an example for a participant who was randomized to have the square be their high-threat CS.

Fig. 2

vmPFC-A in yellow (BAs: 11,24,25); vmPFC-B in red (BAs: 10,11). Regions are apriori anatomical masks derived from the FSL Harvard-Oxford Atlas (Desikan et al., 2006; Goldstein et al., 2007; Makris et al., 2006). Slice coordinates are given in Talairach space.

Fig. 3

a. Baseline (Scan 1) Imaging Results (Aim 1). During the middle phase (6–10 s post stimulus onset), deactivation for high-threat relative to low-threat was observed within the posterior cingulate cortex and precuneus/cuneus (PCC; BAs 29,18,31,7; cluster size 6541 µl) (mean high-threat PSC −0.153, SD 0.187; mean low-threat PSC −0.015, SD 0.176) and in right dorsolateral prefrontal cortex and right middle and superior frontal gyrus (dlPFC; BAs 8,6; cluster size 2457 µl) (mean high-threat PSC −0.081, SD 0.157; mean low-threat PSC 0.014, SD 0.150). Slice coordinates are given in Talairach space. b. Box plots depicting PSC for high-threat (HT), low-threat (LT) and the difference between them (HT-LT) for the dlPFC and PCC during the middle phase for all participants at Scan 1. The center line depicts the median. Circles represent outliers with values between 1.5 and 3 times the interquartile range. c. Mean impulse response function graphs for PCC for all participants at Scan 1. d. Mean impulse response function graphs for dlPFC for all participants at Scan 1.

Fig. 4

a. Box plots depicting PSC for high-threat (HT), low-threat (LT) and the difference between them (HT-LT) for the percent signal change in left ventromedial PFC during the early phase (vmPFC-AL; yellow region in Fig. 2) for all participants at Scan 1. The center line depicts the median. Circles represent outliers with values between 1.5 and 3 times the interquartile range. Stars represent outliers with values more than 3 times the interquartile range. b. Mean impulse response function graphs for vmPFC-AL for all participants at Scan 1.

Fig. 5

This figure depicts effects of prazosin on the subjective experience of anxiety to high-threat during the anticipatory anxiety task. Statistical values for an ANOVA and regression are presented (see Table 2).

Fig. 6

This figure depicts drinking trajectories in percent days abstinent over six weeks, to explore the significant interaction effect for condition (prazosin vs. placebo)*subjective anxiety ratings (Anx) during high-threat. High versus Low Anx is defined by a median split.

Fig. 7

a. This figure depicts drinking trajectories in percent days abstinent over six weeks, to explore the significant interaction effect for condition (prazosin vs. placebo)*percent signal change in left ventromedial PFC during the early phase (vmAL; yellow region in Fig. 2) during high-threat relative to low-threat stimuli. High versus Low vmAL is defined by a median split. b. Mean impulse response function graphs for vmPFC-AL for the 17 participants with the highest vmPFC-AL PSC values at Scan 1. c. Mean impulse response function graphs for vmPFC-AL for the 17 participants with the lowest vmPFC-AL PSC values at Scan 1.