Stress reduction correlates with structural changes in the amygdala

Abstract

Stress has significant adverse effects on health and is a risk factor for many illnesses. Neurobiological studies have implicated the amygdala as a brain structure crucial in stress responses. Whereas hyperactive amygdala function is often observed during stress conditions, cross-sectional reports of differences in gray matter structure have been less consistent. We conducted a longitudinal MRI study to investigate the relationship between changes in perceived stress with changes in amygdala gray matter density following a stress-reduction intervention. Stressed but otherwise healthy individuals (N = 26) participated in an 8-week mindfulness-based stress reduction intervention. Perceived stress was rated on the perceived stress scale (PSS) and anatomical MR images were acquired pre- and post-intervention. PSS change was used as the predictive regressor for changes in gray matter density within the bilateral amygdalae. Following the intervention, participants reported significantly reduced perceived stress. Reductions in perceived stress correlated positively with decreases in right basolateral amygdala gray matter density. Whereas prior studies found gray matter modifications resulting from acquisition of abstract information, motor and language skills, this study demonstrates that neuroplastic changes are associated with improvements in a psychological state variable.

INTRODUCTION

Acute stress initiates hormonal and behavioral responses that enable an organism to make adaptations to environmental demands (Chrousos, 2000). The amygdala has been implicated in both human and animal studies as playing a crucial role during stress responses, including the detection of stressful and threatening stimuli and the initiation of adaptive coping responses (LeDoux, 2000; Hasler et al., 2007). Amygdala-dependent cognition is facilitated during stressful conditions—a useful function for fear-related learning (Shors and Mathew, 1998; Sapolsky, 2003). However, prolonged exposure to stress increases the risk of being affected by a number of mental and physical illnesses (Johnson et al., 1992; Chrousos, 2000; Sapolsky, Romero, & Munck, 2000).

Aberrant amygdala function has been consistently demonstrated across several stress-related psychopathologies. For example, exaggerated amygdala activation has been found in trait anxiety (Stein et al., 2007), post-traumatic stress disorder (PTSD; Rauch et al., 2000; Shin et al., 2004, 2005), social phobia (Birbaumer et al., 1998; Evans et al., 2008; Phan et al., 2006), depression (Drevets et al., 1992; Abercrombie et al., 1998; Sheline et al., 2001; Siegle et al., 2002; Dougherty et al., 2004) and impulsive aggression (Coccaro et al., 2007).

Reports of differences in gray matter structure of the amygdala in pathologic stress conditions have been less consistent (Drevets et al., 2008). While some studies found enlarged amygdala volumes in subjects with affective disorders (Altshuler et al., 1998; Strakowski et al., 1999; Frodl et al., 2002; Lange and Irle, 2004; Weniger et al., 2006), others did not find altered volumes or reported volume reductions (Sheline et al., 1998; Mervaala et al., 2000; Frodl et al., 2003; Frodl et al., 2008). Amygdala findings for patients suffering from PTSD and other anxiety disorders have also been mixed (Gurvits et al., 1996; De Bellis et al., 2000; Gilbertson et al., 2002; Massana et al., 2003; Siegle et al., 2003; Wignall et al., 2004; Milham et al., 2005; Karl et al., 2006; Atmaca et al., 2008; Woon and Hedges, 2008; Hayano et al., 2009). One study with healthy individuals failed to find a correlation between chronic life stress and gray matter volume in the amygdala (Gianaros et al., 2007). These inconsistencies in the literature might result from a number of factors that can impact gray matter measures, such as gender (Wilke et al., 2007), genetics (Meyer-Lindenberg et al., 2006) and volumetry method (Doty et al., 2008).

In contrast to studies of humans, the stress literature with animals is more consistent. Several studies have shown that prolonged stress exposure leads to increases in measures of amygdala structure in rodents (Vyas et al., 2002, 2003; Mitra et al., 2005). Increased dendritic length and increased arborization were reported within the basolateral complex of the amygdala and in the extended amygdala as a result of exposure to chronic immobilization stress (Vyas et al., 2002, 2003). Differences between the results from the human and animal studies might be due to methodological differences. First, the human studies have often investigated amygdaloid volume using MRI, while animal studies have used invasive techniques to look at specific cellular changes within this structure. Second, while most human studies have been cross-sectional investigations of pathologic conditions, the animal studies have been longitudinal, with presumably healthy animals undergoing a controlled chronic stress manipulation. While individual differences are difficult to control and can confound findings in cross-sectional studies, in longitudinal studies these variables remain constant, allowing researchers to selectively vary the factor of interest. However, to our knowledge, no longitudinal neuroimaging studies have examined the influence of stress on amygdala morphology in healthy human beings.

Here, we report a longitudinal MRI study in humans that investigated the correlation between changes in perceived stress and changes in amygdaloid gray matter density following a stress-reduction intervention. Mindfulness-based stress reduction (MBSR; Kabat-Zinn, 1990) is a popular 8-week program developed to help individuals reduce their stress levels and increase psychological well-being. Mindfulness is defined as the non-judgmental awareness of present moment experiences (Kabat-Zinn, 1990). Participants practise meditation techniques designed to increase awareness of present moment experiences such as thoughts, emotions and physical sensations. They also learn to use this awareness in responding more skillfully to stress in their everyday lives. Numerous studies have demonstrated the efficacy of this program in reducing subjective reports of stress and increasing well-being (e.g. Chang et al., 2004; Carmody and Baer, 2008). However, the underlying neural mechanisms of these changes are largely unknown. Since the amygdala has been repeatedly shown to be involved in, and responsive to, an individual’s experience of stress, we hypothesized that changes in perceived stress would be associated with changes in amygdala gray matter density. Correlations within the whole brain were also explored on an exploratory basis.

METHODS AND MATERIALS

Twenty-seven participants (41% males; mean age 35.2 years; SD 6.7 years) who reported high levels of stress during the previous month were enrolled in the study. Individuals were eligible if their score on the perceived stress scale (PSS; Cohen and Williamson, 1988) was ≥1 SD above the population mean. The PSS is a validated self-report questionnaire widely used for assessing an individual’s self-perception of stress. The PSS has 14-, 10- and 4-item versions and has been shown to yield adequate reliability and validity (Cohen et al., 1983; Cohen and Williamson, 1988). In this study, the 4-item version was used to screen potential subjects while the 10-item version was used to assess change in perceived stress before and after the training. Participants gave their responses on a 5-point Likert scale, ranging from never (0) to very often (4). Inclusion criteria was based on the population means according to Cohen et al. (1983; Cohen and Williamson, 1988), namely 4.2 (SD 2.8) for females and 4.7 (SD 3.1) for males.

Further exclusion criteria were: current psychiatric illness or medical illness, ineligibility for MRI scanning (claustrophobia, metallic implants, pregnancy, etc.), or significant previous meditation or yoga experience. The protocol was approved by the Massachusetts General Hospital Institutional Review Board. Written informed consent was obtained from all study participants and they were compensated for completion of assessment procedures.

All participants completed the 8-week MBSR program, consisting of weekly group meetings and daily home mindfulness practises, including sitting meditation and yoga. The sample described here includes participants from two similar studies that both assess the effect of MBSR on brain structure. Sixteen participants received the standard MBSR class held at the Center for Mindfulness at the University of Massachusetts Medical School. Eleven subjects received a shorter version of the MBSR course held at Massachusetts General Hospital that consisted of only 12 contact hours (versus the standard 23 h) and 20 min daily homework practise (versus the standard 40 min). The intervention has been comprehensively described elsewhere (Kabat-Zinn, 1990). Classes took place between April 2005 and June 2008 and were led by several instructors. One enrolled participant was excluded from the data analyses due to non-adherence to home practise requirements (<4 h total of home practise). Data from 26 healthy, right-handed individuals (44% males; mean age 35.7 years, SD 6.3 years) were included in the analyses. Home practise logs demonstrated that participants reported an average of 19.77 h (SD 6.53 h) of prescribed out-of-class mindfulness practise over the 8-week study period. To test whether the amount of practise had an influence on the improvement in stress, a Pearson correlation between the number of hours of mindfulness home practise and the change in PSS scores was performed in SPSS (‘Statistical Package for Social Sciences, Release 12.0.2.’, 2004).

Participants were scanned at the Martinos Center for Biomedical Imaging in Charlestown, MA. Pre-intervention scans were acquired approximately 1 week before the intervention began and post-intervention scans were acquired within the 2 weeks following the intervention. High-resolution MRI data were acquired with a Siemens Magnetom Avanto 1.5 T scanner, using a T1-weighted, magnetization-prepared rapid acquisition gradient echo (MP-RAGE) sequence, consisting of 128 sagittal slices (voxel size: 1.0 × 1.0 × 1.3 mm, TI = 1000 ms; TE = 3.39 ms; TR = 2730 ms; flip angle 7° and matrix 256 × 256 mm).

Anatomical MR images were compared for differences in gray matter density using voxel-based morphometry (VBM; Gaser, 2008), within the SPM5 neuroimaging statistical software (www.fil.ion.ucl.ac.uk/spm/software/spm5/) based in MATLAB 7.1, release 14 (Mathworks Inc., Natick, MA, USA). VBM permits an automated voxel-wise whole-brain statistical comparison of MRI scans. Images were first manually aligned to the anterior commissure after which gray matter, white matter and cerebral spinal fluid components were segmented within native space. We analyzed unmodulated images, which contain the probability within each voxel for being gray matter, i.e. the proportion of gray matter to other tissue types within a region (Good et al., 2001). For each individual, the gray matter segmentations of the post-intervention time-point were co-registered to the image of the pre-intervention time-point. The normalization parameters were calculated for the pre-intervention image only and then applied to the post-intervention image to make sure that regional differences between the images were not removed because of scan-specific normalization. Images were smoothed at 8-mm full width at half maximum with an Isotropic Gaussian Kernel.

Improvement in PSS (post-intervention score minus pre-intervention score; where negative values indicate decreases in PSS scores and positive values indicate increases) was used as the predictive regressor for changes in gray matter density (post-intervention image minus pre-intervention image; where negative values represent a decrease in gray matter density and positive values indicate increases) in a regression analysis. The significance threshold was defined as P < 0.05, corrected for multiple comparisons (false discovery rate) within the search region (height threshold = 0.01, uncorrected). The region of interest was defined as the bilateral amygdalae, according to Tzourio-Mazoyer et al. (2002). Exploratory correlation with gray matter density in the whole brain was performed at a significance threshold of P < 0.01 (uncorrected, 10 voxels).

RESULTS

PSS scores decreased pre- (mean 20.7; SD 5.6) to post-intervention (mean 15.2; SD 4.7; T = 3.7; df = 25; P < 0.001), indicating that the participants benefited from the course. The internal consistency of the PSS was high at both the pre- and post-intervention evaluation (Cronbach’s-α values 0.85 and 0.81, respectively), confirming an adequate reliability of the scale.

To assess whether the amount of individual meditation home practise predicted the improvement in stress, the number of minutes of meditation practise that participants reported on daily logs was correlated with the magnitude of their reduction in stress. With a Pearson correlation coefficient of r = 0.35, the amount of training was mildly correlated with the improvement in stress, though this correlation did not reach statistical significance (P = 0.079; df = 25).

Pre- to post-intervention analyses of the MRI data in SPM confirmed a correlation between change in PSS scores and change in gray matter density within the right amygdala (cluster size: 10 voxels, MNI coordinates of peak voxel x = 32, y = 0, z = −26; voxel-level T = 3.18; P = 0.042, multiple comparisons correction within the amygdala search territory; Figure 1). Larger decreases in perceived stress were associated with larger decreases in amygdaloid gray matter density. The identified region appears to be located in basolateral/lateral regions of the amygdala, based on the atlas by Mai et al.(1997). The correlation of the change in perceived stress and amygdala gray matter density within the left amygdala was not significant.

I'll help you find a way out of suffering from Stress, Crisis, Loneliness, Burnout, Phobia, Conflict, Domestic Violence, Anxiety- and Panic Disorder, Body Illness, generally, whatever the case may be.