Stress-associated neurobiological activity associates with the risk for and timing of subsequent Takotsubo syndrome

Abstract

Aims

Activity in the amygdala, a brain centre involved in the perception of and response to stressors, associates with: (i) heightened sympathetic nervous system and inflammatory output and (ii) risk of cardiovascular disease. We hypothesized that the amygdalar activity (AmygA) ratio is heightened among individuals who develop Takotsubo syndrome (TTS), a heart failure syndrome often triggered by acute stress. We tested the hypotheses that (i) heightened AmygA precedes development of TTS and (ii) those with the highest AmygA develop the syndrome earliest.

Methods and results

Individuals (N=104, median age 67.5 years, 72% female, 86% with malignancy) who underwent clinical 18 F-FDG-PET/CT imaging were retrospectively identified: 41 who subsequently developed TTS and 63 matched controls (median follow-up 2.5 years after imaging). AmygA was measured using validated methods. Individuals with (vs. without) subsequent TTS had higher baseline AmygA (P=0.038) after adjusting for TTS risk factors. Further, AmygA associated with the risk for subsequent TTS after adjustment for risk factors [standardized hazard ratio (95% confidence interval): 1.643 (1.189, 2.270), P=0.003]. Among the subset of individuals who developed TTS, those with the highest AmygA (>mean + 1 SD) developed TTS ∼2 years earlier after imaging vs. those with lower AmygA (P=0.028).

Conclusion

Higher AmygA associates with an increased risk for TTS among a retrospective population with a high rate of malignancy. This heightened neurobiological activity is present years before the onset of TTS and may impact the timing of the syndrome. Accordingly, heightened stress-associated neural activity may represent a therapeutic target to reduce stress-related diseases, including TTS.

Graphical Abstract

Relationship between stress-associated neurobiological activity on ¹⁸F-FDG-PET/CT imaging and risk for subsequent Takotsubo syndrome. ¹⁸F-FDG-PET/CT, ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography; TL, temporal lobe; TTS Takotsubo syndrome; vmPFC, ventromedial prefrontal cortex.

See page 1909 for the editorial comment on this article (doi: 10.1093/eurheartj/ehab026)

Introduction

Takotsubo syndrome (TTS) is an acute, usually reversible heart failure syndrome that is often triggered by acute emotional or physical stressors.¹ Although an underlying link between the brain and heart has long been proposed as a critical factor in the development of TTS,^{2  ,  3} the underlying mechanisms require further clarification.

The response to stressors, such as those that could trigger TTS, is governed by an ensemble of neural structures within the limbic system, which notably includes the amygdala.⁴ Heightened activity within such stress-associated neural tissues augments sympathetic nervous system (SNS) activity, increases leucopoietic activity, and has recently been shown to contribute to the pathogenesis of multiple chronic stress-related diseases, including atherosclerosis.^4–8 Such alterations in stress-associated neural activity, along with the downstream physiologic consequences, have likewise been hypothesized to play a role in TTS pathogenesis.^{2  ,  6  ,  9  ,  10} Within such a construct, chronically heightened stress-associated neural activity (including the amygdala as a key component) may hypothetically prime an individual to mount a more vigorous neurophysiologic response to subsequent stressors, thereby increasing TTS risk. In support of this hypothesis, functional magnetic resonance imaging (fMRI) studies have demonstrated altered neural connectivity in several stress-associated limbic brain regions (i.e. amygdala, anterior cingulate cortex, and hippocampus) among individuals with prior TTS.^{2  ,  3  ,  9} However, it remains unknown whether the observed alterations in neural activity predate TTS, or if they develop as a consequence of the syndrome. Accordingly, further study is needed to assess whether heightened stress-associated neural activity exists prior to the development of TTS and thus may represent a risk factor for the syndrome.

¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG-PET/CT) allows the assessment of stress-associated neurobiological activity, such as amygdalar activity (AmygA; a ratio of amygdalar metabolic activity divided by that of regulatory brain regions). AmygA is elevated in the setting of increased perceived stress and chronic stress conditions, is stable for up to 6 months, and associates with an increased risk of subsequent cardiovascular disease (CVD).^{4  ,  5  ,  11} Moreover, ¹⁸F-FDG-PET/CT imaging enables simultaneous assessment of extra-neural tissues, including bone marrow (BM) activity (an index of leucopoiesis), which partially mediates the relationship between increased AmygA and CVD.⁴ Thus, ¹⁸F-FDG-PET/CT provides a unique opportunity to evaluate the roles of neural and BM activities in linking stress to downstream TTS.

Accordingly, we leveraged ¹⁸F-FDG-PET/CT imaging in a retrospective cohort of individuals who underwent clinically indicated ¹⁸F-FDG-PET/CT imaging prior to TTS onset to test our hypotheses that (i) individuals who develop TTS have heightened AmygA prior to TTS onset, and (ii) among those with subsequent TTS, higher AmygA associates with a shorter duration between imaging and TTS.

Methods

Study design and population

The Partners HealthCare System Research Patient Data Registry (RPDR) was used to identify patients and controls for this retrospective case–control study. The RPDR is a centralized clinical data warehouse for >4.5 million patients from Massachusetts General Hospital (MGH) and Brigham and Women’s Hospital, as well as affiliated community and specialty hospitals in the Boston area (Figure 1). The data registry includes structured and unstructured electronic medical record information, including socio-demographic data, vital signs, laboratory and test results, problem list entries, prescribed medications, billing codes, and clinical notes for healthcare services provided within the system. Using the RPDR online query tool, we first identified individuals with a diagnosis of TTS who had undergone PET/CT imaging at MGH (Boston, MA, USA) between 2005 and 2019, by utilizing the diagnostic terms: ‘Takotsubo cardiomyopathy’, ‘Takotsubo syndrome’, ‘stress-induced cardiomyopathy’, ‘transient left ventricular apical ballooning syndrome’, and ‘reversible left ventricular dysfunction following sudden emotional stress’ among individuals who had undergone clinically indicated whole-body ¹⁸F-FDG-PET/CT imaging. The initial search identified 190 individuals. Next, individuals were manually excluded if (i) TTS occurred prior to imaging, (ii) they had any of the following clinical findings prior to imaging: acute atherosclerotic myocardial infarction, ischaemic cardiomyopathy, decompensated congestive heart failure, severe left-sided valvular heart disease, hypertrophic cardiomyopathy, intracranial tumours, posterior reversible encephalopathy syndrome, history of craniotomy, prior stroke or seizure disorder, or (iii) they had inadequate ¹⁸F-FDG-PET/CT imaging to measure AmygA. Further, two cardiologists blinded to imaging data (MTO and AT) adjudicated all potential cases of TTS using European Society of Cardiology–Heart Failure Association criteria using the available clinical data.¹² If a diagnosis of TTS was not supported by those criteria after a detailed review, potential cases were excluded. The remaining cases were retained and were classified as probable, possible, or less likely TTS (Supplementary material online, Table S1A) to acknowledge uncertainty related to the retrospective nature of the evaluation and the fact that not all subjects had important clinical data (e.g. coronary angiography or follow-up functional assessment). Patients who met diagnostic criteria with a high clinical likelihood of TTS were classified as probable, those who met criteria for TTS with an alternative diagnosis that could not be fully excluded but could potentially explain the presentation were classified as possible, and patients who met criteria for TTS but carried another diagnosis that likely could have explained the presentation were classified as less likely. Of 190 TTS cases initially identified by the data search, ultimately 41 met inclusion criteria.

Figure 1.

Consort diagram for the selection of cases and controls. ESC-HFA, European Society of Cardiology-Heart Failure Association; FDG PET/CT, fluorodeoxyglucose positron emission tomography/computed tomography; HCM, hypertrophic cardiomyopathy; MGH, Massachusetts General Hospital; RPDR Research Patient Data Registry; TTS, Takotsubo syndrome.

Next, controls were matched to TTS cases in three steps. First, utilizing the RPDR online query tool, an initial group of 1368 individuals was identified via an automated matching algorithm. To do so, TTS patients were matched to cases using age, sex, race, and ‘comparative health’. The comparative health measure, which is provided by the RPDR tool, utilizes a weighted sum of all diagnoses, procedures, medications, and health history as a proxy for healthcare usage. Using the RPDR query, 1368 potential controls were identified. Then, from within this group, controls were manually matched to TTS subjects (up to 3:1) by matching oncologic history (presence or absence of malignancy, malignancy type, stage, and treatment). A cohort of 133 potential controls was identified. Of this group, 31 were excluded as they met the pre-defined exclusion criteria (as used for the TTS group) and 39 individuals lacked coverage of the amygdala, resulting in a final cohort of 63 controls. The study complies with the Declaration of Helsinki. The Partners Human Research Committee approved its protocol.

¹⁸F-FDG-PET/CT imaging protocol and measurement of amygdalar and bone marrow activity

¹⁸F-FDG-PET/CT imaging was performed using an integrated scanner (e.g. Biograph 64 Siemens Healthcare, Erlangen, Germany, or similar) in the resting state. Intravenous ¹⁸F-FDG (∼370 MBq) was given following an overnight fast. Three-dimensional PET imaging was performed after ∼1 h of quiet waiting. A non-gated, non-contrast CT was acquired for attenuation correction.

Analysis of stress-associated neural activity was performed in a retrospective analysis using validated research methods by investigators blinded to clinical data.^{4  ,  5} Amygdalar ¹⁸F-FDG uptake (Figure 2A) was measured by placing circular regions of interest (ROIs) over the right and left amygdalae and measuring tracer accumulation as a standardized uptake value (SUV). The bilateral amygdalar mean SUVs were assessed both as an average and individually by dividing by regulatory regional neural uptake to generate a ratio of amygdalar to regulatory activity.¹³ The primary measure for AmygA was derived by averaging the SUVs of the right and left amygdalae, and dividing that value by the mean temporal lobe (TL) SUV.¹³ Additionally, because the ventromedial prefrontal cortex (vmPFC) serves as a regulatory centre under stressful conditions, AmygA was also adjusted for mean vmPFC SUV in a secondary analysis.¹⁴ The inter- and intra-reader variability of AmygA (as the ratio of amygdalar to TL activity) were assessed on a random sample of 30 ¹⁸F-FDG-PET/CT images derived from the current cohort. The absolute mean differences (standard deviation [SD]) in AmygA were 0.07 (0.04) and 0.05 (0.07), for inter- and intra-reader assessments, respectively. The absolute per cent differences (SD) in AmygA were 6.88 (3.95) and 5.20 (6.03) for inter- and intra-reader assessments, respectively. All ¹⁸F-FDG measurements were made using the axial, sagittal, and coronal planes. Additional details are provided in Supplementary material online.

Figure 2.

Amygdalar activity among subjects with and without subsequent Takotsubo syndrome. (A) Axial ¹⁸F-FDG-PET/CT of the amygdala shows increased uptake in a patient with (right) vs. lower uptake in a patient without (left) subsequent TTS. Amygdalar activity is adjusted for regulatory TL or ventromedial prefrontal cortex (vmPFC) activity. (B) Between-group differences in amygdalar activity in models adjusted for TTS risk factors are shown. Individual values are shown. Mean and 95% confidence intervals (error bars) are depicted. SUV, standardized uptake value; TL, temporal lobe; TTS, Takotsubo syndrome; vmPFC, ventromedial prefrontal cortex.

BM ¹⁸F-FDG uptake, a measure of leucopoietic activity, was measured by placing ROIs over the axial sections of each vertebrae from T1 to L5. The mean of the maximum SUV from each vertebra was calculated.⁴

Statistical analyses

Statistical analyses were performed using SPSS version 26 (IBM Corporation, Armonk, NY, USA). Continuous variables are listed as mean and SD, or as median and interquartile range [interquartile range (IQR), 25th–75th percentile] when skewed. Independent sample t-tests were used to compare continuous variables between groups. Categorical variables were compared using Chi-square or Fisher’s exact tests as appropriate. Effect sizes were evaluated with Cohen’s d tests. Logistic regression was employed to evaluate odds ratios (ORs), as exponential β with 95% confidence intervals (CIs), between baseline variables and subsequent TTS. Linear regression was used to identify factors associated with AmygA and test the association between higher AmygA (defined as ≥mean AmygA + 1 SD) with timing of TTS (among those with subsequent TTS) and BM activity. Cox regression was used to calculate hazard ratios (HRs) for TTS events. Backward selection was used where appropriate. Additionally, post hoc stratified subset sensitivity analyses were performed. A two-sided P-value <0.05 was considered significant. Additional details about covariable selection for multivariable models are provided in Supplementary material online.

Results

Study population

Overall, 104 individuals were studied, including 41 cases and 63 controls (median age 67.5 years [IQR: 59.5–74.0], 72% female, 86% with malignancy). Individuals with TTS and controls had similar baseline characteristics, except autoimmune diseases were more common in individuals with TTS (Table 1). Among TTS patients and controls, the interval between baseline imaging and TTS onset, last follow-up, or death was 2.5 [0.5–5.3] years (median [IQR]). Among the subgroup of individuals with subsequent TTS, the interval between imaging and TTS was 0.9 [0.1–4.1] years. Among the subgroup of controls, the interval between imaging and last Follow-up, or death was 2.9 [1.0–6.0] years. Among all individuals, female sex, diabetes, and chronic kidney disease are associated with higher AmygA (Table 2). Among the 41 TTS cases, four patients had recurrent TTS. Additional clinical details, including an expanded description of the TTS cases, can be found in Supplementary material online, Tables S1A–C.

Table 1.

Baseline characteristics

Variable	Overall (n = 104)	Cases (n = 41)	Controls (n = 63)	P-value
Demographics
Median age, years (IQR)	67.5 (59.5–74)	66 (59–74.5)	68 (61–74)	0.983
Female sex	75 (72.1)	29 (70.7)	46 (73)	0.800
Race				0.826
Caucasian	93 (89.4)	37 (90.2)	56 (88.9)
African American	5 (4.8)	2 (4.9)	3 (5.2)
Asian	6 (5.8)	2 (4.9)	4 (6.9)
Vital status
Death during follow-up	48 (46.2)	22 (53.7)	26 (41.3)	0.216
Cardiovascular risk factors and medications
Hypertension	73 (70.2)	29 (70.7)	44 (69.8)	0.923
Hyperlipidaemia	54 (51.9)	22 (53.7)	32 (50.8)	0.775
Current smoking	13 (12.5)	8 (19.5)	5 (7.9)	0.081
Former smoker	59 (56.7)	27 (65.9)	32 (50.8)	0.130
Diabetes	20 (19.2)	6 (14.6)	14 (22.2)	0.337
Aspirin	37 (35.6)	14 (34.1)	23 (36.5)	0.806
Statins
Any	41 (39.4)	15 (36.6)	26 (41.3)	0.633
High intensity	9 (8.7)	3 (7.3)	6 (9.5)	0.696
Beta blockers	29 (27.9)	12 (29.3)	17 (27)	0.800
ACE inhibitors/ARBs	36 (34.6)	11 (26.8)	25 (39.7)	0.178
Diuretics	19 (18.3)	7 (17.1)	12 (19)	0.799
Cardiovascular diseases
Coronary artery disease	10 (9.6)	3 (7.3)	7 (11.1)	0.521
Malignancy history
Prior malignancy	89 (85.6)	35 (85.4)	54 (85.7)	0.961
Prior chemotherapy/radiation	48 (46.2)	16 (39)	32 (50.8)	0.239
Chemotherapy/radiation at the time of imaging	11 (10.6)	2 (4.9)	9 (14.3)	0.127
Cardiotoxic drugs (anthracyclines or trastuzumab)	17 (16.3)	5 (12.2)	12 (19)	0.356
Median time interval between cancer diagnosis and imaging date, years (IQR)	1.8 (0.3, 5.4)	3.2 (0.4, 5.7)	1.4 (0.3, 5.1)	0.366
Psychiatric history
History of depression/anxiety	43 (41.3)	18 (43.9)	25 (39.7)	0.669
Antidepressant drugs	36 (34.6)	18 (43.9)	18 (28.6)	0.108
Antipsychotic drugs	6 (5.8)	4 (9.8)	2 (3.2)	0.160
Other diseases
COPD/asthma	31 (29.8)	16 (39)	15 (23.8)	0.097
Chronic kidney disease	18 (17.3)	8 (19.5)	10 (15.9)	0.632
Autoimmune disease	16 (15.4)	12 (29.3)	4 (6.3)	0.002
Chronic immunosuppressive drug therapy	9 (8.7)	4 (9.8)	5 (8.1)	0.515

Values are shown as n (%), unless noted.

ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; COPD, chronic obstructive pulmonary disease; IQR, interquartile range.

Table 2.

Univariable predictors of Takotsubo syndrome and amygdalar activity

Predictors	Logistic Regression		Linear Regression
	Takotsubo syndrome		Amygdalar Activity
			Amygdalar activity divided by temporal lobe activity		Amygdalar activity divided by medial prefrontal cortex activity
	OR (95%CI)	P-value	Standardized β (95% CI)	P-value	Standardized β (95% CI)	P-value
Age (years)	1.000   (0.969, 1.031)	0.983	0.001   (−0.015, 0.017)	0.903	−0.007   (−0.024, 0.01)	0.403
Female sex	1.120   (0.468, 2.680)	0.800	0.726   (0.309, 1.143)	0.001	0.573   (0.128, 1.018)	0.012
Race	1.156   (0.316, 4.229)	0.826	−0.407   (−1.045, 0.230)	0.208	−0.711   (−1.365, −0.058)	0.033
Cardiovascular risk factors
Hypertension	1.044   (0.441, 2.470)	0.923	0.161   (−0.270, 0.592)	0.460	−0.091   (−0.553, 0.371)	0.696
Hyperlipidaemia	1.122   (0.510, 2.466)	0.775	0.230   (−0.163, 0.622)	0.249	0.363   (−0.056, 0.782)	0.089
Current smoking	2.812   (0.850, 9.302)	0.090	0.168   (−0.429, 0.764)	0.578	0.281   (−0.360, 0.923)	0.386
Former smoker	1.868   (0.829, 4.212)	0.132	−0.005   (−0.403, 0.394)	0.982	0.280   (−0.143, 0.704)	0.192
Diabetes	0.600   (0.210, 1.715)	0.340	0.513   (0.022, 1.004)	0.041	0.457   (−0.051, 0.966)	0.077
Aspirin	0.902   (0.395, 2.057)	0.806	0.060   (−0.353, 0.472)	0.775	−0.005   (−0.455, 0.444)	0.981
Statin therapy	0.821   (0.365, 1.845)	0.633	0.132   (−0.271, 0.536)	0.517	0.282   (−0.151, 0.715)	0.199
Beta blockers	1.120   (0.468, 2.680)	0.800	0.241   (−0.197, 0.679)	0.278	0.304   (−0.170, 0.778)	0.206
ACE inhibitors/ARBs	0.557   (0.237, 1.311)	0.181	0.161   (−0.253, 0.575)	0.443	−0.055   (−0.501, 0.391)	0.807
Diuretics	0.875   (0.313, 2.447)	0.799	0.065   (−0.447, 0.576)	0.803	0.336   (−0.242, 0.914)	0.251
Psychiatric history
History of depression/anxiety	1.190   (0.536, 2.639)	0.669	0.009   (−0.392, 0.410)	0.965	0.326   (−0.104, 0.756)	0.135
Antidepressant drug use	1.957   (0.858, 4.461)	0.110	0.105   (−0.310, 0.520)	0.617	0.350   (−0.098, 0.797)	0.124
Antipsychotic drug use	3.297   (0.575, 18.896)	0.180	0.054   (−0.793, 0.902)	0.899	0.268   (−0.575, 1.111)	0.529
Malignancy history
Prior malignancy	0.972   (0.318, 2.971)	0.961	−0.147   (−0.708, 0.415)	0.606	0.015   (−0.606, 0.636)	0.963
Prior chemotherapy/radiation	0.620   (0.279, 1.378)	0.241	0.028   (−0.368, 0.424)	0.888	−0.273   (−0.697, 0.151)	0.204
Current chemotherapy/radiation	0.308   (0.063, 1.504)	0.145	0.156   (−0.485, 0.798)	0.630	−0.575   (−1.306, 0.156)	0.121
Anthracycline or trastuzumab exposure	0.590   (0.191, 1.822)	0.359	−0.031   (−0.565, 0.504)	0.909	−0.280   (−0.921, 0.362)	0.388
Time interval between cancer diagnosis and imaging date (years)	1.031   (0.964, 1.102)	0.369	−0.019   (−0.053, 0.015)	0.266	−0.025   (−0.062, 0.013)	0.190
Other diseases
COPD/asthma	2.048   (0.872, 4.812)	0.100	−0.097   (−0.529, 0.334)	0.656	0.154   (−0.302, 0.610)	0.505
Chronic kidney disease	1.285   (0.460, 3.586)	0.632	0.925   (0.436, 1.415)	<0.001	0.746   (0.230, 1.261)	0.005
Autoimmune disease	6.103   (1.810, 20.586)	0.004	−0.183   (−0.729, 0.364)	0.509	−0.065   (−0.685, 0.556)	0.836
Chronic immunosuppressive drug therapy	1.232   (0.311, 4.891)	0.766	0.602   (−0.094, 1.298)	0.089	0.415   (−0.324, 1.154)	0.268

Bold values indicate P < 0.05. ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; CI, confidence interval; COPD, chronic obstructive pulmonary disease; OR, odds ratio.

Amygdalar activity in individuals with Takotsubo syndrome vs. controls

Among the clinical and demographic factors assessed, TTS is associated with a history of autoimmune disease (Table 2). In a model adjusted for TTS risk factors, AmygA adjusted for TL activity was higher among individuals who developed TTS vs. controls (mean$\hspace{0.17em}\pm \hspace{0.17em}$SD: 0.98 $\pm$ 0.10 vs. 0.95 $\pm$ 0.07, P = 0.038, Figure 2, Table 3). In analyses wherein amygdalar SUV was adjusted for vmPFC activity, we also observed a significant between-group difference ([Amygdalar SUV÷vmPFC SUV$]\pm$SD: 0.97 $\pm$ 0.10 vs. 0.91 $\pm$ 0.10 for those with subsequent TTS vs. controls, P = 0.012, Figure 2, Table 3). The effect size of the ratio of amygdalar to regulatory activity between cases and controls was 0.37 and 0.53 when TL and vmPFC were used as adjustments, respectively.

Table 3.

Amygdalar activity and subsequent Takotsubo syndrome

Covariables added to the model	Linear regression				Cox regression
	Amygdalar activity ÷ temporal lobe activity		Amygdalar activity ÷ medial prefrontal cortex activity		Amygdalar activity ÷ temporal lobe activity		Amygdalar activity ÷ medial prefrontal cortex activity
	Standardized ß (95% CI)	P-value	Standardized ß (95% CI)	P-value	Standardized HR (95% CI)	P-value	Standardized HR (95% CI)	P-value
Univariable	0.370   (−0.028, 0.768)	0.068	0.535   (0.117, 0.953)	0.013	1.493   (1.090, 2.045)	0.013	1.580   (1.176, 2.125)	0.002
Takotsubo risk factorsa	0.431   (0.024, 0.838)	0.038	0.544   (0.121, 0.968)	0.012	1.643   (1.189, 2.270)	0.003	1.708   (1.240, 2.352)	0.001
Heart failure risk factorsb	0.412   (0.021, 0.803)	0.039	0.583   (0.158, 1.008)	0.008	1.483   (1.072, 2.053)	0.017	1.580   (1.176, 2.125)	0.002
CVD risk factorsc	0.405   (0.015, 0.795)	0.042	0.567   (0.151, 0.983)	0.008	1.483   (1.072, 2.053)	0.017	1.580   (1.176, 2.125)	0.002
Malignancy historyd	0.495   (0.023, 0.967)	0.040	0.640   (0.089, 1.190)	0.024	1.823   (1.254, 2.652)	0.002	1.711   (1.197, 2.445)	0.003
Characteristic that differed between cases and controls (i.e. autoimmune disease)	0.429   (0.031, 0.828)	0.035	0.574   (0.149, 0.998)	0.009	1.643   (1.189, 2.270)	0.003	1.708   (1.240, 2.352)	0.001
Characteristics that associated with amygdalar activity (see Table 2)	0.362   (−0.009, 0.733)	0.056	0.498   (0.104, 0.892)	0.014	1.493   (1.090, 2.045)	0.013	1.580   (1.176, 2.125)	0.002
Antidepressant drugs	0.338   (−0.048, 0.724)	0.085	0.475   (0.063, 0.887)	0.024	1.493   (1.090, 2.045)	0.013	1.580   (1.176, 2.125)	0.002

All multivariable models included age and sex. Bold values indicate P < 0.05.

^aTakotsubo risk factors: hyperlipidaemia, autoimmune disorders, smoking, depression/anxiety, and malignancy.

^bHeart failure risk factors: hypertension, coronary artery disease, diabetes, and smoking.

^cCVD risk factors: hypertension, hyperlipidaemia, diabetes, and smoking.

^dMalignancy history: prior malignancy, prior chemotherapy/radiation, chemotherapy/radiation at the time of imaging, cardiotoxic drugs (anthracyclines or trastuzumab), and time interval between cancer diagnosis and imaging date.

CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio.

Amygdalar activity vs. Takotsubo risk

Two separate approaches were employed to evaluate associations between baseline AmygA and risk of TTS. First, in adjusted logistic regression analyses, higher AmygA associated with greater odds of developing TTS [standardized OR (95% CI) 1.64 (1.03, 2.61), P = 0.036] and remained significant in multivariable models, stratified Subgroup analyses, and when vmPFC was employed as the regulatory tissue correction of AmygA (Supplementary material online, Tables S2 and S3).

Secondly, time-adjusted Cox regression was performed. In that analysis, AmygA independently predicted subsequent TTS [standardized HR (95% CI) 1.643 (1.189, 2.270), P = 0.003] after adjustment for TTS risk factors. That is, each SD increase in AmygA is associated with a 64% increase in TTS risk. This association remained significant through other multivariable adjustments (Table 3), separate assessments of unilateral AmygA (Supplementary material online, Table S4), when vmPFC was used as the regulatory tissue correction of AmygA (Table 3), and in sensitivity analyses with follow-up limited to 5 years (Supplementary material online, Table S5).

Amygdalar activity vs. timing of Takotsubo

We additionally tested whether, among the subset of individuals with subsequent TTS, higher AmygA (i.e. ≥ mean + 1 SD) was associated with a shorter interval between imaging and TTS onset. Individuals with higher AmygA developed TTS ∼2 years before those with lower AmygA [β (95% CI) −2.72 (−5.12, −0.32), P = 0.028, Supplementary material online, Figure S1] after adjusting for TTS risk factors, remaining significant after alternatively adjusting for CVD (P = 0.029) and heart failure risk factors (P = 0.043).

Amygdalar activity vs. leucopoietic activity

AmygA (≥ vs. <mean + 1 SD) associated with BM (i.e. leucopoietic) activity [standardized β (95% CI) 0.65 (0.09, 1.21), P = 0.024, Supplementary material online, Figure S2] and remained significant after adjustment for age and sex (P = 0.038) and CVD risk factors (P = 0.027). Notably, BM activity did not associate with TTS (P = 0.640).

Discussion

This study is the first to employ ¹⁸F-FDG-PET/CT imaging to demonstrate that heightened stress-associated neurobiological activity precedes the development of TTS among a population with a high rate of malignancy. Furthermore, we observed that, among individuals who developed TTS, AmygA associated with the timing of subsequent TTS. We also identified a significant relationship between AmygA and leucopoietic activity. Together, the findings of the current study thus provide insights into one potential mechanism that may contribute to the ‘heart–brain connection’ in TTS by suggesting that chronically heightened stress-associated neurobiological activity (i.e. the ratio of amygdalar to regulatory activity) may potentially impact both the risk for and timing of subsequent TTS.

Amygdalar activity independently associates with subsequent Takotsubo syndrome

Patients with TTS commonly identify an acute stressor as the precipitating factor of the syndrome; thus, a mechanism linking the brain to the heart has long been suspected of playing a key role in its pathogenesis.^{2  ,  3} A recent fMRI study showed that individuals with prior TTS manifest abnormal neural connectivity among regions associated with autonomic function and limbic system regulation (including the amygdala).¹⁵ However, those neural alterations were observed in individuals with prior TTS, raising the key question of whether such changes were the cause or result of TTS. This study findings provide important answers to that question by demonstrating that individuals who subsequently develop TTS manifest increased AmygA years before developing TTS. Such heightened AmygA may predispose individuals to TTS by potentiating the sympathetic, neurohormonal, and inflammatory consequences of future stressors.

Mechanistic insights

While a consensus on the pathophysiology of TTS is lacking, it is likely that stressors trigger a complex pathophysiologic cascade,¹⁶ including catecholamine toxicity, abnormal myocardial perfusion, myocardial stunning, and endothelial dysfunction.¹⁷ Several stress-associated neural centres in the limbic system (including the amygdala) play a critical role in responding to stressors and may initiate this cascade. Limbic system activation induces hormonal, autonomic, and immune changes associated with fear and stress.^{18  ,  19} The amygdala, via its efferents, effects physiologic changes through the SNS.^{6  ,  15} A perceived threat activates noradrenergic neurons in the brainstem and the sympatho-adrenomedullary axis, causing the release of catecholamines, likely contributing to TTS.⁶ These study findings suggest that increased AmygA predates the development of TTS and may effectively prime the limbic system to respond more vigorously to acute stressors, thus increasing TTS risk. The relationship between AmygA and timing of TTS (among TTS cases) provides further support for this priming hypothesis. That higher relative AmygA associates with earlier TTS development suggests that those with the highest AmygA are relatively more prone to develop TTS upon stress exposure and may develop the syndrome sooner.

Additionally, our results demonstrate an association between amygdalar and leucopoietic activities. This well-described association points to a possible neuro-immune mechanism.^{4  ,  5} In mice, stress, through sympathetic signals, triggers leucopoietic activity that accelerates innate immune cell output and cytokine production.⁷ In humans, individuals with heightened AmygA have increased leucopoietic activity that also links to atherosclerosis and CVD events, thus delineating a neuro-immune link to CVD.^{4  ,  5} In concert with this hypothesized neuro-immune mechanism, TTS involves myocardial macrophage infiltration, changes in the distribution of monocyte subsets, and increased systemic pro-inflammatory cytokines.²⁰ Importantly, we did not observe a significant association between BM activity and TTS risk. Furthermore, consistent with prior studies,^21–23 we observed a significant association between autoimmune disorders and TTS, suggesting a possible link between inflammation and TTS. Thus, additional work is needed to assess the potential relationship between heightened systemic inflammation (as well as leucopoietic activity) and TTS risk. Nevertheless, the relationship between AmygA and TTS remained robust to adjustment for autoimmune diseases and immunosuppressive medications in multivariable models. Taken together, these study results support a concept wherein individuals with chronically heightened stress-associated neural activity are predisposed to an exaggerated systemic response to stress that may prompt the development of TTS (Figure 3).

Figure 3.

Hypothesized neurobiological pathway linking stress to Takotsubo syndrome. Individuals with a higher ratio of amygdalar to regulatory activity are primed to have a heightened response upon encountering subsequent acute stressors, resulting in greater activation of the sympathetic and immune system and increased Takotsubo risk.

Potential role of other brain regions

Assessment of AmygA involves dividing amygdalar metabolic activity by the activity of regulatory brain regions. This ratio (amygdalar to regulatory neural activity) is designed to reflect an interaction between functionally connected stress-associated neural regions that promote vs. regulate the stress response. The amygdala promotes a greater stress response, while regulatory regions (e.g. vmPFC) attenuate this response.²⁴ In this study, we pre-specified the use of TL activity for background adjustment.⁴ However, we alternatively adjusted for vmPFC activity, given its important role in reducing the stress response¹⁴ and recent findings of altered prefrontal connectivity among individuals with TTS.⁴ Notably, using vmPFC activity to adjust for AmygA yielded even more robust relationships between stress-associated neural activity and TTS. These findings are consistent with the theory that the amygdala participates within a network of stress-responsive neural tissues, and that the balance of activity within that network (e.g. heightened AmygA and/or lowered regulatory activity) influences the risk for adverse physiological consequences that result from stress. This construct is also consistent with observations of hypo-connected sub-networks involving the amygdala³ and decreased vmPFC blood flow and connectivity among individuals with prior TTS.⁹ Additional brain regions (e.g. insular cortex, anterior cingulate cortex, and hippocampus) that were not evaluated in this study are involved in regulating the response to stress and have altered connectivity in individuals with prior TTS.^{2  ,  3} Accordingly, while the ratio of amygdalar to regulatory activity appears to have a role in the risk for developing TTS, additional study is needed to clarify the roles of other important neural centres in TTS development.

Implications

Previous research has shown that increased AmygA may participate in the pathology of a number of stress-related cardiovascular and metabolic diseases.⁴ This study provides important insights into the link between AmygA and TTS that may improve the care of at-risk patients. Importantly, the ratio of amygdalar to regulatory activity may represent a useful target for both pharmacologic and non-pharmacologic behavioural interventions (e.g. stress reduction) in individuals at high risk for TTS and other stress-associated diseases. Because recurrent TTS is associated with considerable morbidity and mortality,¹ heightened AmygA may be a particularly useful therapeutic target for secondary prevention. The downstream components of this pathway may also represent targets to attenuate TTS risk by using anti-inflammatory drugs to reduce heightened inflammation or using beta-blockers to reduce SNS discharge (though beta-blockers have not proven beneficial in TTS to date).²⁵

Limitations

This study was a modestly sized single-centre retrospective case–control study and is subject to the inherent limitations. The population largely consisted of individuals with a diagnosis of malignancy (86%), a known TTS risk factor, which may limit the generalizability of our findings. Nevertheless, the controls in this study were carefully matched by type and stage of malignancy as well as malignancy treatment to those who developed TTS, which would serve to offset the frequency of malignancy among the TTS cases and address our inability to directly and uniformly measure cancer activity on ¹⁸F-FDG-PET/CT imaging. While imaging performed in the context of a possible or known diagnosis of malignancy is likely to induce stress, ¹⁸F-FDG-PET/CT imaging was performed under resting conditions in this study, and all individuals were subject to the same conditions. The results were robust to adjustment for clinically diagnosed depression and anxiety, but we were unable to assess perceived stress or evaluate for subclinical psychiatric symptoms in this retrospective study. As such, the relationship between subclinical stress conditions and AmygA may be incompletely evaluated. While our findings support the presence of chronically increased AmygA on resting imaging years before the development, we were unable to directly measure instantaneous changes in brain activity or metabolism in response to the acute stress during the TTS event and, therefore, could not directly assess a causal relationship. Accordingly, a subsequent study could also implement fMRI in conjunction with ¹⁸F-FDG-PET in a broader population to clarify the role of the stress response in triggering TTS. Due to limitations of clinical whole-body ¹⁸F-FDG-PET/CT imaging, we were unable to study all brain regions involved in the stress response, and there may have been a small degree of spillover in measurements from structures neighbouring the amygdala. These other important brain regions (e.g. insular cortex, anterior cingulate cortex, and hippocampus) should be evaluated in future PET/MRI studies. Finally, the study is potentially limited by ascertainment bias due to the use of diagnostic codes to identify cases of a rare disease in a retrospective cohort with strict inclusion criteria. To minimize this impact, each patient’s chart was carefully adjudicated by two blinded cardiologists; however, broader study in a general population is needed for confirmation. Nevertheless, these limitations are counterbalanced by unique and important insights provided by imaging performed before the development of TTS. Furthermore, given the rarity of TTS, this study would be extremely challenging to perform prospectively, and the fact that significant differences were still identified in a small study supports AmygA as a meaningful marker for TTS risk.

Conclusions

Among a population of individuals with a high frequency of malignancy, those who subsequently develop TTS have higher baseline ratio of amygdalar to regulatory activity compared with those who do not. This heightened stress-associated neurobiological activity is present years before disease onset and may represent a previously unrecognized TTS risk factor. Among individuals who later developed TTS, higher AmygA was associated with a shorter interval between imaging and development of TTS. Collectively, these findings shed light on a ‘heart–brain connection’ representing a neurobiological mechanism that contributes to TTS and suggests potential targets to reduce TTS risk.²⁶

Supplementary material

Supplementary material is available at European Heart Journal online.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Funding

This work was supported by National Institutes of Health (NIH) [T32HL076136, KL2TR002542, and K23HL151909] and American Heart Association [18CDA34110366] [M.T.O.], NIH P01HL131478, R33HL141047, R01HL149516, R01HL152957, R56AR077187, and R01HL137913 as well as support from the Harvard Medical School Osher Center [A.T.], and NIH [R01HL137562, R01HL130539, and K24HL113128] as well as support from A. Curtis Greer and Pamela Kohlberg [T.G.N.].

Conflict of interest: A.T. received institutional grants from Genentech for research outside the submitted work. T.G.N. reports acting as a consultant for Parexel, Bristol Myers-Squibb, Aprea Therapeutics, and Intrinsic Imaging unrelated to the current research. M.T.O. reports acting as a consultant for Intrinsic Imaging unrelated to the current research. The remaining authors have no disclosures.