Frontostriatal brain activation is associated with the longitudinal progression of cardiometabolic risk

Abstract

Objective:

Cardiometabolic risk refers to a set of interconnected factors of vascular and metabolic origin associated with both cardiovascular disease and various brain disorders. While midlife cardiometabolic risk is associated with future brain dysfunction, emerging evidence suggests that alterations in autonomic and central nervous system function may precede increases in cardiometabolic risk.

Methods:

The present study tested whether patterns of cerebral blood flow in brain areas associated with autonomic regulation were associated with increases in overall cardiometabolic risk. A community sample of 109 adults with resting systolic blood pressure between 120–139, diastolic blood pressure between 80–89, or both, underwent pseudocontinuous arterial spin labeling to quantify cerebral blood flow responses to cognitively challenging tasks. Cardiometabolic risk and cerebral blood flow measurements were collected at baseline and at a two-year follow-up.

Results:

Regression analyses showed that greater frontostriatal cerebral blood flow responses to cognitive challenge were associated with higher cardiometabolic risk at follow-up (β = 0.26 [95% C.I. = 0.07, 0.44], t = 2.81, p = 0.006, ΔR² = 0.04). These findings were specific to frontostriatal brain regions, as frontoparietal, insular-subcortical, and total cerebral blood flow did were not associated with progression of cardiometabolic risk. Moreover, cardiometabolic risk was not associated with frontostriatal cerebral blood flow responses two years later.

Conclusions:

Frontostriatal brain function may precede and possibly forecast the progression of cardiometabolic risk.

Cardiometabolic risk refers to a set of interconnected factors of metabolic origin (e.g., atherogenic dyslipidemia, elevated blood pressure, elevated plasma glucose, and adiposity).^1–3 Elevated cardiometabolic risk is predictive of diabetes,⁴ cardiovascular disease,⁵ and a variety of brain disorders.^6,7 While there is a well-established link between midlife cardiometabolic risk and brain dysfunction,⁸ mounting evidence suggests that alterations in nervous system function may precede or co-occur with increases in cardiometabolic risk.^9–14 Hence, individual differences in brain functionality may predict future cardiometabolic risk.

A possible basis for a link between brain functionality and later cardiometabolic risk is dysregulated central control of the cardiovascular system. For example, blood pressure is regulated in part through autonomic innervation of the heart and blood vessels.¹⁵ Epidemiological studies have shown that individual differences in levels of sympathetic and parasympathetic control of the heart prospectively predict increases in blood pressure.^11,12 With respect to brain function, a recent study showed that adults with aberrant patterns of cerebral blood flow in brain regions associated with central control of blood pressure during a cognitively challenging task were more likely to show increases in blood pressure two years later.¹⁵

Another possible basis by which brain functionality may be associated with later cardiometabolic risk is via dysregulated central control of the liver. For example, sympathetic nervous system activation via stimulation of the splanchnic nerves increases plasma glucose via gluconeogenesis.¹⁶ Conversely, hypoglycemia increases sympathetic outflow to the liver.¹⁷ Lipoprotein metabolism is also regulated via bidirectional autonomic pathways between the hypothalamus and liver.¹⁸ While we do not know of any prospective cardiometabolic studies linking autonomic control of the liver with future metabolic risk, there are a number of cross-sectional and longitudinal studies showing that individual differences in autonomic control of the heart are associated with hyperglycemia.¹⁹ The latter findings are consistent with the notion that dysregulated central control of autonomic outflow to the peripheral organs may link brain functionality with later cardiometabolic risk. However, it is not currently known if elevated cardiometabolic risk is explained by the function of brain networks for autonomic control.

The autonomic nervous system innervates all major organ systems (e.g., cardiovascular, digestive, and endocrine) and is centrally regulated by cortical & subcortical brain regions.^20–23 Previous human neuroimaging studies suggest that both the structure and function of brain regions involved in autonomic control are correlated with cardiometabolic risk factors.^8,10,24–28 Though cardiometabolic risk may only be associated with brain atrophy late in life,⁸ a recent study from our research group showed that adults with aberrant patterns of cerebral blood flow in frontostriatal brain areas (i.e., ventral striatum, ventral prefrontal, & subgenual anterior cingulate) during cognitively challenging tasks was associated with increases in blood pressure two years later.¹⁰ Moreover, recent prospective studies show resting metabolic activity in the amygdala predict both increased adiposity²⁷ and incident diabetes.²⁸ While many of these studies examined only one aspect of cardiometabolic risk, these findings suggest the previously untested and broader possibility that patterns of brain activity may predict increases in overall cardiometabolic risk.

Here, we first aimed to establish whether patterns of cerebral blood flow in frontostriatal brain areas predictive of increases in blood pressure are also associated with increases in overall cardiometabolic risk. If so, then the findings would suggest that the way in which the brain responds to cognitive challenges is associated in some way with future increases in cardiometabolic risk. Second, to test the specificity of frontostriatal brain regions in relation to cardiometabolic risk, we examined whether cardiometabolic risk is associated with patterns of cerebral blood flow in other brain regions associated with autonomic responding and the central regulation of blood pressure. Our third aim was to test the alternative direction of effects (i.e., cardiometabolic risk → brain function). The results of our third aim was meant to test whether the association between cardiometabolic risk and brain function is unidirectional or bidirectional within a longitudinal study design framework.

METHODS

Participants.

Participants were recruited using community advertisements and databases of individuals interested in participation in research, e.g., those participating in the Adult Health and Behavior Study.²⁹ Participants were screened via telephone with the goal of enrolling healthy adults in a longitudinal study on the impact of possible increases in blood pressure over two years. Inclusion criteria were age (35 to 65 years) and either normotensive levels of blood pressure or prehypertensive levels of blood pressure (either systolic 120–139 mmHg or diastolic 80–89 mmHg, and not taking any cardiovascular medications). Exclusion criteria included: a) general medical conditions: pregnancy, ischemic coronary artery disease (prior positive stress test, myocardial infarction or revascularization), cancer, chronic liver disease, chronic kidney disease, or type 1 diabetes; b) neuropsychiatric conditions: stroke, multiple sclerosis, serious head injury, epilepsy, brain tumor or major mental illness (e.g., bipolar disorder and schizophrenia); and c) use of prescription medications for hypertension or psychotropic drugs. Additional exclusions included: current heavy alcohol consumption (≥ 21 standard drinks per week, non-fluency in English (speaking/reading English every day for <10 years), night shift workers, and magnetic resonance imaging scanning incompatibility due to claustrophobia, metal in body, or body habitus. The University of Pittsburgh Institutional Review Board approved the study and all participants gave informed consent.

Data collection took place between August 2010 to May 2015. A total of 219 individuals (49 normotensive, 170 prehypertensive) completed the cardiometabolic risk assessment during the baseline portion of the study. The mean and standard deviation of cardiometabolic risk measures collected on this total sample were used to create z-scores that were then combined to define cardiometabolic risk. A sample of 113 prehypertensive participants were retested two years later. Normotensive participants were not followed. Four participants did not complete the neuroimaging component of the study at baseline which resulted in 109 participants with complete baseline data. Participant characteristics for those with complete data are shown in Table 1. An additional 10 participants did not complete the neuroimaging session at follow-up, resulting in 99 participants available for testing the association between cardiometabolic risk the frontostriatal response at follow-up.

Table 1:

Descriptive statistics for baseline participant characteristics (n = 109)

	Mean (SD)
Age	48.56 (6.97)
Sex (% female)	46.79%
Race (% white)	62.39%
Years of education	15.13 (2.50)
Systolic blood pressure (mmHg)	123.70 (6.91)
Diastolic Blood Pressure (mmHg)	81.47 (4.91)
Body Mass Index (kg/m2)	29.48 (5.66)
Waist circumference (cm)	99.18 (12.86)
High density lipoprotein (mg/dl)	53.02 (14.10)
Triglyceride (mg/dl)	120.13 (60.04)
Glucose (mg/dl)	93.94 (26.48)

Measurement of Cardiometabolic Risk.

For resting blood pressure determinations, participants were asked to abstain from eating, smoking, drinking caffeinated beverages, and heavy physical activity for at least 30 minutes prior. Participants were seated with back supported and feet flat on the floor for at least 5 minutes. Cuff size was selected based upon arm circumference. Systolic and diastolic (5th phase) blood pressure was determined twice separated by two minutes using a 2–3 mmHg/sec deflation rate. Two readings at each of two such sessions were averaged. All measurements were obtained by staff specifically trained in blood pressure measurement using published guidelines.³⁰

Height and weight were measured with participants’ shoes removed. Height and weight were measured digitally using a SECA electronic model #242 Digital output (Hamburg, Germany) and a DETECTO model 758C Digital scale (Webb City, MO), respectively. Body mass index was calculated as: weight (kg) / height (m) ². Waist circumference was measured following exhalation centered at the umbilicus. Glucose, triglyceride, and high density lipoprotein levels were measured from a blood draw following a 12-hour fast. Assays were done by the Heinz Chemistry and Nutrition Laboratory of the University of Pittsburgh Graduate School of Public Health.

The cardiometabolic risk is based upon five factors as defined by international guidelines: waist circumference, systolic and diastolic blood pressure, and fasting serum glucose, triglycerides and high density lipoprotein cholesterol.³¹ As done previously by us and others, waist circumference, systolic and diastolic blood pressure, log glucose, log triglyceride and high density lipoprotein (reverse scored) were converted to standardized z-scores.³² Glucose and triglycerides were log transformed to normalize the distribution. High density lipoprotein was reverse scored. Systolic and diastolic blood pressure were averaged together after standardization. These standardized scores were then combined using unit weighting. Cardiometabolic risk was calculated in the same fashion at the two-year follow-up, using the mean and standard deviation from the baseline sample to calculate the z-scores.

Measurement of Cerebral Blood Flow.

Participants practiced each of the tasks prior to a 1-hour neuroimaging session. The magnetic resonance imaging scanner was a 3T Trio TIM whole-body scanner (Siemens) equipped with a 12-channel phased-array head coil. The neuroimaging session included structural images: T1- weighted image, fast spin echo T2-weighted sequence, and a fast fluid-attenuated inversion recovery. T1-weighted images were obtained using a 3-dimensional fast-gradient MPRAGE (magnetization prepared rapid gradient-echo) sequence: repetition time 1,540 ms, echo time 3.04 ms, 192 sagittal slices, flip angle 8°, field of view 256 mm, slice thickness 1 mm with no gap, voxel dimensions 1 mm³. The MPRAGE images were used for normalization.

Pseudocontinuous arterial spin labeling (pCASL) was used to collected perfusion images with (label) and without (control) labeling of the arterial blood.³³ These scans were collected using gradient echo-planar imaging (interleaved ascending, matrix 64 × 64, 3.75 mm in plane resolution, 6-mm-thick slices, 2790 Hz/pixel, echo time 18 ms, flip angle 90°, repetition time 3.6 seconds). The pCASL labeling period was 1.5 seconds with a 1.1-second transit delay and 1-second acquisition for 20 slices. The pCASL label was placed 9 cm below the center of the imaging volume. A coronal phase-contrast magnetic resonance angiogram (single 80-mm-thick slice, flip angle 25°, repetition time 50 ms, matrix: 256 × 256) was used to ensure the pCASL plane was placed orthogonally to the internal carotids and vertebral arteries. Label/control measurements were acquired for 12 minutes (100 pairs of images).

Preprocessing of the perfusion images included realignment to the first image of the series, coregistration to each participant’s MPRAGE image, spatial normalization to the International Consortium for Brain Mapping 152 template (Montreal Neurological Institute), and resliced to an isotropic voxel size of 3 mm³. Following smoothing with a 12-mm fullwidth at half-maximum isotropic gaussian kernel, perfusion images were created by subtracting the unlabeled images from the labeled images. Cerebral blood flow images were created using the tracer kinetic model.³⁴ Voxels with negative perfusion values and values > 2.5 * median were set to missing.

Separate mL/100 resting and task voxel wise cerebral blood flow images and total cerebral blood flow values, both in units of g per minute, were created by averaging across task blocks. The block design included 3 minutes of a control phase preceding and following 6 minutes of the active phase. The initial active task was 2-back working memory task that included the presentation of a sequence of letters, one at a time. In this task, the participant pressed a button every time a presented letter matched the letter presented 2 spots back in the sequence. The control task, a 0-back, required the participant to press a button every time the presented letter matched a target letter specified prior to the beginning of the task.

The second task was a modified version of the multisource interference test (MSIT).^35,36 For the active task, numbers 1, 2, and 3 were matched to the second, third, and fourth finger positions on a response glove. Three numbers were presented evenly spaced across the screen for each trial; 2 were identical (e.g., 3 3 1). The participant was required to use the finger corresponding to the differing number to execute a button press. Interference arises in the task when the spatial location of the non-identical number presented on the screen does not match the numeric mapping of the fingers. During the control task, the location of the differing number on the screen always matched the numeric mapping of the fingers.

Perfusion images were preprocessed using Statistical Parametric Mapping software (SPM8; Wellcome Trust Centre for Neuroimaging) and a customized ASL toolbox.³⁷ General linear models with a weighted least squares (SPM toolbox, RobustWLS, http://www.diedrichsenlab.org/imaging/robustWLS.html) were used to quantify voxelwise task effects on cerebral blood flow using boxcar functions to model task conditions; the CBF changes from control to task phase were evaluated from preplanned contrast analyses.

We then obtained estimates of change in cerebral blood flow in 16 bilateral regions of interest by extracting the first principal component of the contrast data in each of the regions across all participants. Selected regions were the amygdala, Brodmann area 6 (BA6, prefrontal cortex), dorsolateral prefrontal cortex, midcingulate cortex, ventrolateral prefrontal cortex, ventromedial prefrontal cortex, anterior insula, cerebellum, dorsal anterior cingulate cortex, dorsal striatum, pons, posterior cingulate cortex, posterior parietal, subgenual anterior cingulate cortex, thalamus, and ventral striatum. These regions were selected based on prior literature in which blood pressure or blood pressure reactivity during cognitive tasks was related to brain structure or function.^38–45

Preliminary analyses of cerebral blood flow responses in our selected regions for a previous paper showed substantial correlations between areas, suggesting the existence of underlying components.¹⁵ To account for the correlation between regions and to reduce the chances of type 1 error, we conducted a principal component analysis with varimax rotation which showed 3 components with eigenvalues > 1. Component scores were then generated by summing regional values for areas that loaded > .48 on each component.

Data Analytic Plan

All analyses were conducted in R version 3.6.1 using the jtools package to test our regression models and for creating the tables and figures. First, cardiometabolic risk at the two-year follow-up was regressed onto cardiometabolic risk at baseline (to account for individual differences in initial cardiometabolic risk) and each cerebral blood flow composite at baseline. The standardized regression coefficient for frontostriatal cerebral blood flow at baseline was used to test the hypothesis that higher frontostriatal activity at baseline was associated with increased cardiometabolic risk at the two-year follow-up. Because our previous report using the same dataset showed frontostriatal activity was associated with increased blood pressure, we calculated cardiometabolic risk without blood pressure and tested the association with frontostriatal activity to see if our findings with frontostriatal activity were dependent on blood pressure.

The specificity of frontostriatal brain regions in relation to cardiometabolic risk was tested by examining the standardized regression coefficient for frontoparietal and insular-subcortical regional cerebral blood flow at baseline. Finally, frontostriatal cerebral blood flow at the two-year follow-up was regressed onto frontostriatal cerebral blood flow at baseline (to account for individual differences in initial frontostriatal responding) and cardiometabolic risk at baseline. The standardized regression coefficient for cardiometabolic risk at baseline was used to test the alternative direction of effects (i.e., cardiometabolic risk → brain function). Each regression model is reported with no additional covariates (unadjusted; model 1), adjusted for total cerebral blood flow at baseline (model 2), and adjusted for age, sex, race (white vs. not white), and years of education (model 3).

Results

Table 1 shows participant characteristics at baseline. Overall, participants were primarily white and had some education past high school. Blood pressure levels were moderately high and most participants were overweight or obese. Tables are also included in the online supplement showing means and standard deviations for individual cardiometabolic risk factors, as well as cerebral blood flow values in individual regions of interest (see Tables S1 – S3, Supplemental Digital Content).

Frontostriatal cerebral blood flow and cardiometabolic risk

Table 2 shows the standardized parameter estimates from the multiple regression model of the prospective relationship between each of the composite cerebral blood flow responses at baseline and cardiometabolic risk at two-year follow-up. In support of our first hypothesis, greater frontostriatal cerebral blood flow responses were associated with higher cardiometabolic risk at follow-up (β = 0.26 [95% C.I. = 0.08, 0.44], t = 2.80, p = 0.006, ΔR² = 0.03; see Figure 1). The pattern of findings was the same before and after adjustment for potential confounding variables. Moreover, the pattern of findings was the same when cardiometabolic risk was calculated without systolic blood pressure (β = 0.19 [95% C.I. = 0.07, 0.31], t = 3.04, p = 0.003, ΔR² = 0.03; see Table 3).

Table 2.

Regression models of the relationship between baseline cerebral blood flow composites and cardiometabolic risk at two-year follow-up

	Model 1	Model 2	Model 3
Baseline Frontostriatal CBF	0.26 **	0.26 **	0.26 **
	[0.08, 0.44]	[0.08, 0.44]	[0.07, 0.45]
Baseline Frontoparietal CBF	0.01	0.01	−0.03
	[−0.16, 0.18]	[−0.17, 0.18]	[−0.22, 0.15]
Baseline Insular-Subcortical CBF	−0.07	−0.07	−0.07
	[−0.25, 0.11]	[−0.25, 0.11]	[−0.25, 0.12]
Baseline Cardiometabolic Risk	0.68 ***	0.68 ***	0.68 ***
	[0.55, 0.81]	[0.54, 0.81]	[0.54, 0.82]
Baseline Total CBF		−0.03	0.01
		[−0.16, 0.11]	[−0.14, 0.15]
Age			−0.01
			[−0.16, 0.14]
Sex			−0.23
			[−0.55, 0.08]
Race			0.01
			[−0.30, 0.32]
Education			0.03
			[−0.13, 0.19]
N	109	109	109
R2	0.54	0.54	0.55

Standardized regression coefficients are reported with the 95% confidence interval below in brackets. Each model is adjusted for cardiometabolic risk at baseline. Model 1 is unadjusted for any further covariates. Model 2 is adjusted for total cerebral blood flow at baseline. Model 3 is also adjusted for age, sex, race, and education. CBF = Cerebral blood flow;

^***p < 0.001;

^**p < 0.01;

^*p < 0.05.

Figure. 1.

Regression slope, 95% confidence interval, and observed data for the relationship between baseline frontostriatal cerebral blood flow response and cardiometabolic risk at two-year follow-up.

Table 3.

Regression models of the relationship between baseline frontostriatal cerebral blood flow and cardiometabolic risk without systolic blood pressure at follow-up

	Model 1	Model 2	Model 3
Baseline Frontostriatal CBF	0.19 **	0.18 **	0.16 *
	[0.07, 0.31]	[0.06, 0.31]	[0.03, 0.28]
Baseline Cardiometabolic Risk w/o Blood Pressure	0.73 ***	0.73 ***	0.72 ***
	[0.61, 0.85]	[0.60, 0.85]	[0.59, 0.85]
Baseline Total CBF		−0.07	−0.03
		[−0.20, 0.05]	[−0.16, 0.10]
Age			0.04
			[−0.09, 0.17]
Sex			−0.25
			[−0.53, 0.02]
Race			−0.06
			[−0.34, 0.22]
Education			0.04
			[−0.10, 0.19]
N	109	109	109
R2	0.59	0.60	0.62

Standardized regression coefficients are reported with the 95% confidence interval below in brackets. Each model is adjusted for cardiometabolic risk at baseline. Model 1 is unadjusted for any further covariates. Model 2 is adjusted for total cerebral blood flow at baseline. Model 3 is also adjusted for age, sex, race, and education. CBF = Cerebral blood flow;

^***p < 0.001;

^**p < 0.01;

^*p < 0.05.

Frontoparietal and insular-subcortical cerebral blood flow

In addition to our frontostriatal composite, we also tested frontoparietal and insular-subcortical cerebral blood flow responses as the predictors of cardiometabolic risk at follow-up (see Table 2). Frontoparietal (β = 0.01 [95% C.I. = −0.16, 0.18], t = 0.14, p = 0.887, ΔR² < 0.01) and insular-subcortical (β = −0.07 [95% C.I. = −0.25, 0.13], t = −0.74, p = 0.457, ΔR² < 0.01) cerebral blood flow responses only accounted for a small amount of variance in cardiometabolic risk. Overall, these tests suggested specificity for the frontostriatal effects.

Is cardiometabolic risk associated with future frontostriatal cerebral blood flow responses?

Our last set of regression models tested the alternative direction of our frontostriatal effects (Table 4). Cardiometabolic risk did not account for a substantial amount of variance in frontostriatal cerebral blood flow at the two-year follow-up (β = 0.06 [95% C.I. = −0.14, 0.25], t = 0.56, p = 0.576, ΔR² < 0.01). Overall, this null finding is incompatible with cardiometabolic risk as a predictor of future frontostriatal cerebral blood flow.

Table 4.

Regression models of the relationship between baseline cardiometabolic risk and frontostriatal cerebral blood flow at two-year follow-up

	Model 1	Model 2
Baseline Cardiometabolic Risk	0.06	0.09
	[−.14, 0.26]	[−0.10, 0.29]
Baseline Frontostriatal CBF	0.21 *	0.15
	[0.01, 0.41]	[−0.05, 0.35]
Age		−0.15
		[−0.36, 0.05]
Sex		−0.13
		[−0.55, 0.28]
Race		0.48 *
		[0.05, 0.92]
Education		0.31 **
		[0.09, 0.54]
N	99	99
R2	0.05	0.17

Standardized regression coefficients are reported with the 95% confidence interval below in brackets. Each model is adjusted for frontostriatal cerebral blood flow at baseline. Model 1 is unadjusted for any further covariates. Model 2 is adjusted for age, sex, race, and education. CBF = Cerebral blood flow;

^***p < 0.001;

^**p < 0.01;

^*p < 0.05.

Discussion

Here, we tested the possibility that activation of frontostriatal brain regions is associated with increases in cardiometabolic risk factors over two years. In a prospective study design, we found that frontostriatal brain activation was associated with future cardiometabolic risk, and this association was not explained by increases in blood pressure alone. These results do not rule out an unknown common (3^rd factor) cause of both frontostriatal activity and cardiometabolic risk, but they do indicate that changes in brain function may precede progression of cardiometabolic risk. We have also shown evidence for the specificity of frontostriatal brain regions, as frontoparietal, insular-subcortical, and total cerebral blood flow were not associated with progression of cardiometabolic risk.

Our results suggest the possibility that neural responses to cognitive challenges may forecast the progression of cardiometabolic risk. These findings extend our previous work showing the same frontostriatal response was associated with increased blood pressure over two years.¹⁰ The predictive value to the progression of both blood pressure and overall cardiometabolic risk is not surprising given that the same brain systems appear to be multiply involved in stress, autonomic, and metabolic control processes.^46,47 Overall, our frontostriatal composite is similar to what has been termed a mesocortical limbic system that has excitatory and inhibitory effects on both motivation/drive and metabolic regulation.^48,49 Activity in this system has been shown to balance effort required to attain reinforcement from the value of the reinforcement.⁵⁰ Activation in prefrontal areas comparable to the regions in our frontostriatal composite have been shown to be predictive of subsequent weight gain in a sample of adolescents.⁵¹ Thus, the predictive value of the frontostriatal activation may be explained by the involvement of these brain regions in the recruitment of motivation/effort during challenging tasks.

Overall, regional cerebral blood flow in frontostriatal brain regions decreased relative to the non-challenging control task. Those with advancing cardiometabolic risk, however, showed relatively increased cerebral blood flow in frontostriatal regions. This pattern of results may suggest relatively greater allocation of effort during task performance to the possible detriment of metabolic regulation. Chronic responses such as this could plausibly relate to cardiometabolic risk. Transient responses within dopamine related areas/mesocortical limbic system are known to chronically remodel patterns of brain responding.^48,50 Our data, however, are unable to precisely support such an interpretation given the complexity of inhibitory and excitatory interactions within the mesocortical limbic system. Future studies employing multiple waves of data should test whether the onset of differences in frontostriatal areas precede the increase in cardiometabolic risk.

Our findings disagree with the conclusions of a recent cross-sectional report that did not find evidence supporting a relationship between cardiometabolic risk variables and patterns of brain activation during stress.⁵² There are a number of study differences that may explain the discrepant findings. These include differences in study design (cross-sectional vs. longitudinal), sample size differences and corresponding statistical power (N = 57 vs. 109), brain imaging methodologies (blood oxygen level dependent signal vs. pseudo continuous arterial spin labelling), and analytical testing (mass univariate voxel-wise testing vs. a priori regions of interest). Whether these methodological differences explain their lack of findings is unknown, but future studies should explicitly attend to patterns of activation in the striatum using multiple neuroimaging and task paradigms.

Our null findings with insular-subcortical brain activation are somewhat in disagreement with recent studies that show resting metabolic activity in the amygdala predict both increased adiposity²⁷ and incident diabetes.²⁸ The most notable study differences include brain imaging methodology (18F-fluorodeoxyglucose positron emission tomography vs. pseudo continuous arterial spin labelling) and outcome measure (individual components vs. global cardiometabolic risk). Future studies should determine the degree to which resting vs. task-induced amygdala activation precede increases in cardiometabolic risk.

While our prospective design has a number of strengths relative to previous cross-sectional neuroimaging studies of cardiometabolic risk,^8,52 reliance on only two-waves of data is a limitation. Multiple time points across the lifespan would provide potentially useful information about the sequence of events that cause individual differences in cardiometabolic risk and frontostriatal responses. One potential model to test in this regard is that stress hormones (e.g., adrenocorticotropic hormone, cortisol) alter frontostriatal pathways involved in regulation of reward that leads to increased intake of highly palatable foods and weight gain.¹³ Such a study would need to be relatively large and begin early in life, as is the case with the Adolescent Brain Cognitive Development study.⁵³

If alteration of frontostriatal brain circuits reliably precede the development of cardiometabolic risk, then targeting modifiable risk factors that negatively impact frontostriatal brain circuitry may be a promising primary prevention or even risk stratification and monitoring strategy. The notion that the brain may play a role in diseases related to cardiometabolic risk is a promising topic for research.

Abbreviations;

BA

Brodmann area

C.I.

confidence interval

CBF

Cerebral blood flow

dL

deciliter

mg

milligram

mL

milliliter

mmHg

millimeters of mercury

MPRAGE

magnetization prepared rapid gradient-echo

ms

millisecond

MSIT

multisource interference test

pCASL

pseudocontinuous arterial spin labeling

SPM

statistical parametric mapping