Neural Correlates of Stress and Abdominal Obesity in Patients with Coronary Artery Disease

Abstract

Objective:

To investigate the relationship between waist circumference as a measure of abdominal obesity and brain responses to stress among patients with coronary artery disease (CAD)

Methods:

Patients with CAD (N=151) underwent acute mental stress tasks in conjunction with High Resolution Positron Emission Tomography (HR-PET) and radiolabelled water imaging of the brain. Brain responses to mental stress were correlated with waist circumference.

Results:

Waist circumference was positively correlated with increased activation in right and left frontal lobes (beta values ranging from 2.81 to 3.75 in the paracentral, medial, and superior gyri), left temporal lobe, left hippocampal, left amygdala, left uncus, and left anterior and posterior cingulate gyri (beta values ranging from 2.93 to 3.55). Waist circumference was also negatively associated with left and right parietal lobe, right superior temporal gyrus, and right insula and precuneus (beta values ranging from 2.82 to 5.20).

Conclusion:

Increased brain activation in the brain regions involved in the stress response and autonomic regulation of the cardiovascular system during psychological stress may underlie stress-induced overeating and abdominal obesity in patients with CAD.

Introduction

Coronary artery disease (CAD) remains a major cause of morbidity and mortality worldwide with a prevalence that is increasing worldwide (1). Excess abdominal fat has shown to be an important risk factor for CAD (2-5). Precise measurements of abdominal fat content requires the use of radiological imaging techniques such as computerized tomography, magnetic resonance imaging or dual energy X-ray absorptiometry (6, 7). However, given that these imaging methods are not practical in large-scale field epidemiology, waist circumference (WC) has often been used as a surrogate to determine abdominal obesity. WC has increased significantly in the United States in both women and men in recent decades, (8). Few treatments for obesity have been shown to result in lasting weight loss (9). Likewise, most prevention programs have resulted in only modest and short-term changes in obesity-related metabolic abnormalities and have not been able to prevent the development of obesity (9, 10). Given the increasing societal burden of abdominal obesity and CAD, understanding the complex mechanisms that drive this condition is necessary in order to optimize and target interventions that could prevent obesity or result in lasting weight loss in obese individuals.

One factor that may play a role in the development and maintenance of obesity is psychological stress. Evidence from population-based and clinical studies have shown a strong association between chronic stress and increased adiposity and weight gain (11-14). A number of potential mechanisms for this have been proposed, including changes in eating patterns and increased food cravings (15-17). These behavioral changes may be driven by alterations in neuropeptides and neurohormones that play a role in both stress and appetite / metabolism (18). Stress is associated with a preferential deposition of abdominal fat (19-22). Stress-related psychiatric disorders have been linked to obesity as evidenced by studies that showed an increase in depression in obese individuals (20) and increased obesity in patients with childhood abuse-related PTSD (23-26). These disorders may be independently associated with increased rates of obesity through adverse lifestyle behaviors, neurohormonal and/or metabolic dysregulation (27-29).Diet can also affect mood, including deficiencies of specific nutrients including Vitamin B12 and folate. (30-35), while diets that are high in saturated fats can increase the risk for CAD.

In addition to diet-related effects, central insulin resistance has been suggested as an important factor contributing to altered motivation for food as well as changes in brain regions involved in reward pathways (36, 37). In obese individuals, insulin levels have been shown to correlate positively with the neural activity in corticolimbic-striatal brain regions during favorite-food and stress cues (16). These data suggest that insulin sensitivity may affect specific human brain responses during exposure to stress. However, the mechanisms by which this occurs are not known

The purpose of the present study was to investigate the relationship between WC as a measure of abdominal obesity and brain responses to stress among patients with CAD. We hypothesized that among patients with CAD, those with increased WC would show increased stress-induced changes in brain areas involved in the regulation of memory, emotion, and/or peripheral cardiovascular responses to stress, including hippocampus, insula, and cingulate gyrus.

Methods

Study Cohort

All participants were participants recruited from the Mental Stress Ischemia Mechanisms and Prognosis Study (MIPS) (38-41). Recruitment methods for MIPS cohort were described elsewhere (41). Briefly, between the years of 2011 and 2014, 695 patients between 30 and 80 years of age with confirmed stable CAD were prospectively enrolled from Emory-affiliated hospitals and clinics. Patients were excluded if they had unstable angina, myocardial infarction, or decompensated heart failure within one week of their baseline visit; if they had any severe comorbidities expected to decrease their life expectancy to the next 5 years; or if they had systolic or diastolic blood pressure above 190 and 115 mmHg, respectively, on the day of testing. Patients were also excluded if they had a history of alcohol or substance abuse in the past year, severe psychiatric disorders other than depression, inflammatory diseases, malignancies, organ transplants, permanent atrial fibrillation, and severe neurological disorders such as dementia or Parkinson’s disease. Pregnant women, patients with chronic steroid use above 1500 meg per day, and patients who were unable to complete the mental stress testing were also excluded from study participation. From this initial cohort of 695 patients, a random sample of 151 patients were selected and underwent brain scans. Data regarding the sociodemographic characteristics, medical history, and medication history of all participants were collected by a research nurse using standard questionnaires, chart reviews and in-person interviews and were verified using medical records. Height, weight, and waist circumference were measured for each individual. These measurements were used to calculate the body mass index (BMI; in kg·m⁻²). All study participants provided informed consent and the study was approved by the Emory University Institutional Review Board.

Mental Stress Testing

All patients underwent mental stress testing using a standardized protocol, as previously described (41, 42). Briefly, following 30 minutes of resting in a quiet room, participants were asked to perform two control tasks, followed by two mentally stressful conditions, each lasting approximately 2 minutes. All participants were scanned twice during each of the 4 testing conditions. During the control conditions, participants were tasked with counting out loud and discussing a neutral event. The stress conditions included arithmetic and public speaking tasks.

The mental arithmetic task included multiplication, division, addition and subtraction. Patients were required to listen to a recorded message that provided instructions for the tasks. All tasks were performed under time pressure and negative feedback regarding the performance and the time spent in the task given. For the mental arithmetic task the level of difficulty was increased until participants were unable to successfully complete the 3 consecutive arithmetic problems. The purpose of calibrating difficulty to individual performance is to have a similar level of stress for all participants regardless of ability. The negative feedback was standardized by having the physician use a stop watch to time the performance and paper and pencil to score performance, and to provide standardized negative feedback. This type of negative feedback has shown to consistently result in an increased stress level. PET data recorded from the mental arithmetic task was compared to neutral counting.

In the public speaking task, two scripted scenarios of stressful interpersonal situations were provided to each participant and they were asked to develop a speech regarding these events. Each participant was given 2 minutes to prepare each speech and 3 minutes to present it in front of an audience and camera. Participants were told that their speeches would later be evaluated. Data from the public speaking tasks was only gathered during the speech. During the neutral recall, each participant was asked to recall a recent neutral memory. PET data recorded from the speech task was compared to neutral recall. These tasks were randomized.

All participants were required to hold beta-adrenergic antagonists and nitrate and calcium channels blockers for a minimum of 24 and 12 hours, respectively, prior to mental stress testing.

Brain imaging during mental stress

High Resolution-Positron Emission Tomography (HR-PET) brain imaging was conducted using the High Resolution Research Tomograph (HRRT) (Siemens, Inc., Knoxville, TN), with a spatial resolution of 2 mm (43). Brain perfusion was investigated by quantifying the regional activation in response to both stressful and control tasks. Ten seconds into each HR-PET scan, 20 mCi of radio-labeled water (H2[O15]), was injected to measure blood perfusion within the brain. Each scan lasted for approximately two minutes and electrocardiogram and vital signs were continuously monitored with physician present. A total of eight brain scans were performed for each individual with two scans during each of the 2 control (counting aloud and recalling a neutral event) and 2 scans for the stress (arithmetic and public speaking) conditions.

Data Analysis

Analysis of HR-PET images was performed using statistical parametric mapping (SPM8) software, as previously described (44, 45). Pre-processing of HR-PET images, including realigning, normalizing, and smoothing images, was completed identical to previous research (44, 46). HR-PET imaging analysis was completed on two levels. The first level consisted of creating individual contrast maps to determine brain areas with increased regional blood flow indicating activation (Blood FlowNet = Blood Flow_{Mentai stress Task} – Blood Flow _{control Task}). These individual contrast maps were then used within the second-level voxel-wise regression analysis (SPM8; www.fil.ion.ucl.ac.uk/spm). Voxel-wise regression models were completed for brain activations and waist circumference with both positive and negative correlations assessed. . In order to average the contrasts, we first computed the contrast between mental arithmetic vs neutral counting, and second between the speech preparation vs neutral memory recall, and subsequently averaged these two contrasts. For the voxel-wise regression analysis, significant cluster peaks were identified using a threshold of p < 0.005 and a minimum cluster size of 11 voxels to balance Type I and II error as had been previously determined (47) Contrast maps were created for second level regression analyses in order to identify areas with increased and decreased blood flow following mental stress. Dice coefficient was computed using a built-in MATLAB function. The Dice coefficient was calculated to quantify the amount of overlap between two stress tasks. This index is calculated as the ratio of twice the intersection of two volumes divided by the sum of the two volumes, where A and B are the two volumes to be compared (DS-Coeffs(A, B) = 2(A ⋂ B)/(A + B)). Values range between 0 and 1 with a value of 1 indicating a total concordance between the two volumes, whereas a value of 0 indicating no overlap between the two volumes.

Differences in demographic, clinical and behavioral variables between BMI groups were assessed using ANOVA for continuous variables and Chi-square tests for categorical variables. All analyses were done using SPSS version 22.

Results

Table 1 presents the demographic characteristics of all 151 participants. Overall, the study sample consisted of 68.2% male with a mean age of 62 ± 8 years old and a mean BMI of 30.17 ± 6.0. The most common comorbidity in the cohort was dyslipidemia (82.1%), followed by hypertension (76.2%), and history of prior myocardial infarction (35.8%). As shown in Table 2, WC was positively correlated with male sex and presence of diabetes. No significant correlation was observed between age and WC (B −0.04, 95% CI (−0.15, 0.05)).

Table 1.

Baseline demographics of the study cohort

	All Patients (N=151)
Demographics
Age, y ± SD	62 (8)
Male, %	103 (68.2)
African American, %	54 (35.8)
Comorbidities,
Diabetes, %	52 (34.4)
Dyslipidemia, %	124 (82.1)
Heart Failure, %	22 (14.6)
Hypertension, %	115 (76.2)
History of myocardial infarction, %	54 (35.8)
Depression, %	79 (52.3)
History of smoking, %	23 (15.2)
Medication Use
ACE Inhibitors, %	68 (44.7)
Antidepressants, %	53 (34.7)
Aspirin, %	132 (87.3)
Beta Blockers, %	111 (73.3)
Diuretics, %	47 (30.7)
Statins, %	136 (90.0)

ACE: Angiotensin converting enzyme

Table 2.

Differences in waist circumference between participants based on clinical and demographic variables.

Variable	Waist Circumference (Mean ± SD)		P- Value
	Absent	Present
Male	39 ± 5	41 ± 5	0.02
African American	41 ± 5	40 ± 5	0.21
Comorbidities
Diabetes	40 ± 4.8	42 ± 6	0.003
Dyslipidemia	40 ± 7	41 ± 5	0.67
Heart Failure	41 ± 5	39 ± 5	0.21
Hypertension	40 ± 6	41 ± 5	0.44
History of myocardial infarction	40 ± 5	41 ± 5	0.70
Depression	40 ± 5	41 ± 6	0.15
History of smoking	41 ± 5	39 ± 5	0.10
Medication Use	Not Taking	Taking
ACE Inhibitors	41.2 ± 5	41.1 ± 5	0.93
Antidepressants	40 ± 5	41 ±6	0.41
Aspirin	42 ± 5	41 ± 5	0.21
Beta Blockers	41.6 ± 5	41.0 ± 6	0.54
Diuretics	41.0 ± 6	41.7 ± 5	0.68
Statins	42.2 ± 6	41.0 ± 5	0.45

Overall, as shown in Figure 1,in response to mental stress testing, large areas of the bilateral frontal lobe including the inferior, medial, and superior gyri, bilateral temporal lobe, parietal lobe, bilateral anterior cingulate, and the bilateral posterior cerebellum showing increased activation. There were large overlaps between the mental arithmetic and public speaking tasks with a dice coefficient of 0.55.

Figure 1:

Overall brain activation for mental arithmetic (levt), and public speaking (right), for both stress-induced activation and deactivation.

As shown in Table 3 and Figure 2 and 3, waist circumference was positively correlated with increased activation in right and left frontal lobes (paracentral, medial, and superior gyri), left temporal lobe, left parahippocampal, left amygdala, left uncus, left limbic lobe, left occipital lobe and left anterior cingulate gyri (Table 3 and Figure 2). WC was negatively associated with right and left insula, left and right parietal lobe, right and left superior temporal gyri, and right and left cerebellum (Table 4, and Figure 4).

Table 3:

Associations between Areas of Increased Activation During Mental Stress Testing and Waist Circumference.

			Talairach
Voxel Number	Brain Region	Brodmann Area	X	Y	Z	Z Score
41	R Frontal Lobe, Paracentral Lobule	31	2	−17	44	3.75
96	L Parahippocampal Gyrus		−34	−18	−14	3.55
	L Temporal Lobe, Sub-Gyral		−30	−31	−4	3.16
	L Parahippocampal Gyrus	36	−42	−20	−11	2.91
32	L Temporal Lobe, Superior Gyrus	22	−51	−56	14	3.38
37	L Limbic Lobe, Posterior Cingulate	29	−8	−50	7	3.38
58	L Limbic Lobe, Uncus		−24	−1	−25	3.38
	L Limbic Lobe, Uncus	38	−16	6	−34	2.96
	L Limbic Lobe, Uncus	36	−20	−2	−32	2.76
12	L Frontal Lobe, Paracentral Lobule	6	−2	−28	55	3.28
11	L Frontal Lobe, Middle Gyrus	11	−34	34	−15	3.26
21	L Amygdala		−30	−5	−15	3.23
31	L Frontal Lobe, Medial Gyrus	10	0	66	−2	3.21
27	L Anterior Cingulate	32	−12	28	22	3.17
13	L Occipital Lobe, Inferior Gyrus	17	−20	−94	−9	3.17
19	L Frontal Lobe, Sub-Gyral		−24	45	5	3.15
32	L Frontal Lobe, Paracentral Lobule	5	−8	−42	57	3.13
23	R Frontal Lobe, Superior Gyrus	9	10	54	34	3.13
11	L Frontal Lobe, Cingulate Gyrus	32	0	23	36	3.10
15	L Temporal Lobe, Middle Gyrus	21	−38	−7	−25	3.07
17	R Frontal Lobe, Medial Gyrus		0	52	1	3.06
11	R Frontal Lobe, Paracentral Lobule	5	10	−44	61	3.04
19	L Frontal Lobe, Middle Gyrus		−26	32	19	2.98
	L Frontal Lobe, Middle Gyrus		−28	37	10	2.68
16	R Frontal Lobe, Superior Gyrus	9	22	48	35	2.97
14	R Parietal Lobe, Superior Lobule	7	22	−55	61	2.93
15	L Frontal Lobe, Medial Gyrus	10	−6	61	12	2.88
20	L Frontal Lobe, Middle Gyrus	8	−26	27	37	2.85
13	L Frontal Lobe, Middle Gyrus	9	−28	23	27	2.81

Brain Regions with significant (p < 0.005) positive relationships between waist circumference (cm) and brain activation as measured with [¹⁵O]H₂O positron emission tomography following mental stress (mental arithmetic, public speaking). The XYZ coordinates identify the cluster peak with sub-cluster peaks occurring > 8mm away also presented. Voxel number refers to number of voxels in cluster with Z-score of cluster/sub-cluster peak also presented.

Figure 2:

Sagittal brain slices showing areas with significant (p < 0.005) positive relationships between waist circumference (cm; red) and brain activation

Sagittal brain slices showing areas with significant (p < 0.005) positive relationships between waist circumference (cm; red) and brain activation as measured with as measured with [¹⁵O]H₂O positron emission tomography following mental stress (mental arithmetic, public speaking). Values underneath brains present Talairach x axis values (negative = left, positive = right hemisphere).

Figure 3:

Positive correlations between brain activations (net difference in blood flow in mL⁻1·min⁻1·100g⁻¹) and waist circumference (cm). Brain activations were computed from mean activity within significant (p < 0.005) clusters as identified with voxel-wise regression.

Table 4:

Associations between Areas of Decreased Activation During Mental Stress Testing and Waist Circumference.

			Talairach
Voxel Number	Brain Region	Brodmann Area	X	Y	Z	Z Score
450	R Insula	13	53	−36	20	5.20
	R Parietal Lobe, Inferior Lobule	40	48	−29	33	4.25
	R Parietal Lobe, Inferior Lobule	40	55	−32	27	4.14
127	R Parietal Lobe, Precuneus	7	12	−70	39	4.32
	R Parietal Lobe, Superior Lobule	7	24	−58	41	3.27
	R Parietal Lobe, Precuneus	7	22	−60	48	2.95
49	R Cerebellum		24	−53	−13	4.29
40	L Parietal Lobe, Precuneus	7	−20	−64	42	4.17
	L Parietal Lobe, Precuneus	7	−12	−56	43	2.88
86	R Cerebellum		53	−48	−21	4.02
40	L Parietal Lobe, Superior Lobule	7	−36	−54	50	3.80
145	R Parietal Lobe, Inferior Lobule	40	40	−54	41	3.73
100	R Temporal Lobe, Fusiform Gyrus	20	57	−34	−18	3.70
	R Temporal Lobe, Middle Gyrus	21	59	−36	−12	3.41
82	R Lentiform Nucleus		18	21	−4	3.69
	R Lentiform Nucleus		24	12	−1	3.18
24	R Parietal Lobe, Precuneus	7	4	−58	38	3.58
30	R Temporal Lobe, Superior Gyrus	22	59	−11	7	3.57
70	L Parietal Lobe, Angular Gyrus	39	−51	−64	37	3.43
	L Parietal Lobe, Inferior Lobule	40	−44	−50	39	3.28
	L Parietal Lobe, Inferior Lobule	40	−53	−54	38	3.12
45	R Temporal Lobe, Superior Gyrus	22	57	2	−3	3.40
12	R Temporal Lobe, Inferior Gyrus	20	67	−17	−23	3.38
19	R Frontal Lobe, Inferior Gyrus	44	59	5	15	3.33
20	R Frontal Lobe, Precentral Gyrus	43	51	−7	14	3.23
15	L Temporal Lobe, Superior Gyrus	22	−63	−31	8	3.22
13	L Parietal Lobe, Precuneus	7	−8	−33	42	3.17
21	L Temporal Lobe, Superior Gyrus	13	−53	−42	22	3.16
13	L Cerebellum		−42	−75	−18	3.13
11	L Insula	13	−44	−3	12	3.13
17	R Parietal Lobe, Inferior Lobule	40	53	−42	46	3.07
14	R Temporal Lobe, Middle Gyrus		59	−56	3	3.05
12	R Temporal Lobe, Fusiform Gyrus	37	50	−63	−15	3.04
12	R Frontal Lobe, Inferior Gyrus	47	34	29	5	2.98
18	R Cerebellum		6	−61	−14	2.98
15	R Occipital Lobe, Lingual Gyrus	18	18	−74	−8	2.95
11	R Temporal Lobe, Fusiform Gyrus	20	55	−21	−26	2.91
12	L Parietal Lobe, Inferior Lobule	40	−55	−41	44	2.88
19	L Occipital Lobe, Cuneus	17	−8	−95	5	2.83
	L Occipital Lobe, Cuneus	18	−14	−101	7	2.82

Brain Regions with significant (p < 0.005) negative relationships between waist circumference (cm) and brain activation as measured with [15O]H2O positron emission tomography following mental stress (mental arithmetic, public speaking). The XYZ coordinates identify the cluster peak with sub-cluster peaks occurring > 8mm away also presented. Voxel number refers to number of voxels in cluster with Z-score of cluster/sub-cluster peak also presented.

Figure 4:

Negative correlations between brain activations (net difference in blood flow in mL⁻1·min⁻1·100g⁻¹) and waist circumference (cm). Brain activations were computed from mean activity within significant (p < 0.005) clusters as identified with voxel-wise regression.

Discussion

The results of the present study revealed that waist circumference in CAD patients were associated with exacerbated neural responses in the left temporal lobe, left occipital lobe, left anterior and posterior cingulate gyrus, right paracentral lobule of the frontal lobe, and increased activation in hippocampus and amygdala. The brain regions found by our analysis have also been shown to be involved in pathways influenced by eating behaviors, regions affected by hyperinsulinemia, and areas controlling the central processing of autonomic function. It is important to point out that in our study links between the identified brain areas and their function were not directly investigated. However, our findings may suggest that differences in brain responsivity to stress could play a role in the link between stress and vulnerability to obesity.

While stress has been associated with obesity and abdominal obesity (11-14, 48), the mechanisms underlying this process are not fully elucidated (49). One of the proposed pathways is through the influence of stress on eating behaviors (50, 51). Previous studies have shown that both acute and chronic stress increase food intake, particularly high-fat and high-sugar foods (52, 53). Food cravings have also shown to partly mediate the relationship between chronic stress and increased BMI (15). Previous studies in participants without CAD have shown higher corticolimbic activations in obese, but not lean, individuals in response to favorite-food cue and stress conditions. These regions included the inferior and superior frontal gyrus, middle temporal gyrus, inferior occipital gyrus, and parahippocampus. In addition, the magnitude of insulin resistance, has shown to positively correlate with the activation of these same regions in response to both favorite-food cue and stress conditions in obese but not lean individuals (16, 54). In our study, we found strong correlations between similar regions that were more activated in response to stress among those with higher measures of abdominal obesity. These findings suggest that among obese individuals, insulin resistance may accentuate stress- and food cue–related neural responses by impairing insulin’s ability to suppress pathways implicated in reward-motivation. Therefore, it is a possibility that in fact hyperinsulinemia is a factor influencing the brain responses to mental stress. Therefore, brain responses to stress are not necessarily the cause of increased abdominal obesity and increases in WC. As our study was cross-sectional in nature, the casual relationship between brain activation and increase in WC could not be clearly investigated.

In addition to hyperinsulinemia, abdominal obesity is highly associated with an increase in inflammatory markers including C-reactive protein and interleukin 6 (55, 56). There is evidence that inflammation causes mood disturbances that correlate with brain activity in anterior cingulate cortex- a brain region implicated in the etiology of depression (57). Brain regions implicated in stressor-evoked blood pressure reactivity have also been shown to mediate the relationship between inflammation and blood pressure in healthy participants (58). These data suggest that it could be abdominal obesity that influences brain responses to stress and not necessarily brain activation that causes changes in WC in response to stress.

Accumulating data from animal and human studies suggest that dysregulation of the autonomic nervous system activity plays a pivotal role in the etiology of metabolic abnormalities including obesity, central obesity and insulin resistance (59, 60). The results from brain-imaging studies in humans have proposed a central autonomic network (CAN) designed to modulate autonomic outputs (61-67). Our results showed that several of the regions identified in the CAN were more activated in patients with increased BMI and waist circumference. These regions included the anterior cingulate, and the paracentral lobe of the frontal lobe which showed higher activation with increases in both BMI and waist circumference. Other regions of the CAN including the hippocampus, and parts of ventromedial prefrontal cortex (Brodmann’s Area 10) were also positively correlated with increases in waist circumference, while activation of the right insula and cerebellum were seen in patients with elevated BMI. These results suggest that altered CAN activity to stress may play an important role in higher susceptibility of patients to obesity.

In the present study, we found strong associations between waist circumference and brain activity in the left hippocampus. Increases in waist circumference, which is a measure of central obesity has shown to better assess body fat distribution in CAD patients. Compared to BMI, measurement of waist circumference is a stronger predictor of future development of CAD (68). Prior studies have also shown that patients with CAD with normal BMI but who are centrally obese have the highest cardiovascular mortality compared with other patterns of adiposity (4, 5). The hippocampus, has been implicated in the regulation of body weight through functions including memory coding and inhibitory processes related to food intake (69). It has been shown that increased activity of the hippocampus correlates with increased emotional and uncontrolled eating (70) and because of its role in memory, the hippocampus may also be an important structure to guide food decisions. Our findings expand upon previous work showing the importance of these structures following an acute stress as a possible neurobiological signature in individuals with increased waist circumference. Future studies are needed to assess whether the observed brain-evoked patterns of activation also predict future outcomes in this vulnerable patient population.

The present study had a number of limitations. First, the cross-sectional nature of the study precluded any assessment of causality. Second, all individuals were selected from a cohort of patients with stable CAD, which may limit the generalizability of our findings to individuals without CAD. Finally, women comprised only 32.3% of the total participants. Therefore, while no sex differences were seen in the patterns of brain activation in response to stress, our sample of women was small. In addition, the PET imaging was acquired during speech delivery which could potentially confound the results. Finally, mental stress testing was only performed during a single day, and therefore we could not ascertain whether the brain responses to stress seen in our study are stable and trait-like throughout life.

In summary, we found that obese individuals with CAD exhibit exaggerated neural responses in regions involved in stress, memory and emotion. These neural responses may underlie eating patterns and food cravings which could subsequently lead to increases in waist circumference. Our findings underscore the importance of better understanding environmental exposures such as psychological stress in the pathophysiology of obesity in order to design more effective treatment and prevention strategies. It is however unclear whether the brain responses to mental stress are similar between patients with CAD and the healthy participants. Future studies are required to explore the similarities and differences in the patterns of brain reactivity to mental stress between papulations with and without coronary disease.