Introduction

Heart failure (HF) is a risk factor for stroke, cognitive impairment, and structural damage to the brain.1 This association is attributed to an impaired systolic output and reduced cerebral blood flow, as well as ischemic damage to the brain structure. Brain injury observed in patients with HF ranges from overt stroke to microinfarction, brain atrophy, and white matter lesions. Patients with HF are affected by cognitive impairment, mood disorders, depression, and memory loss. Several gray matter structures such as the hippocampus, cingulate, cuneus, amygdala, putamen, or thalamus play a crucial role in cognitive functions.2,3

The aim of our study was to perform a comprehensive assessment of left ventricular systolic function as well as brain magnetic resonance imaging, with volumetric evaluation and segmentation of structures implicated in cognitive functions, in a mixed population of patients with impaired left ventricular ejection fraction (EF) and healthy controls.

Material and methods

The study population consisted of a mixture of healthy individuals who self-reported on their health status and patients with impaired left ventricle function. Healthy controls were included in the study if the medical history and physical examination revealed normal resting electrocardiography and there was no noticeable disease. We did not exclude patients with blood pressure less than 140/90 mm Hg on a single measurement. The remaining patients had various degrees of left ventricular systolic dysfunction caused by myocardial infarction or dilated cardiomyopathy and EF higher than 40%. The exclusion criteria included implanted electronic devices, renal failure, cancer, stroke, and current atrial fibrillation.

The final study population consisted of 63 individuals, including 38 healthy controls in group 0 and 25 patients with reduced left ventricular EF in group 1. The University Ethics Committee approved the study protocol and written informed consent was obtained from all participants.

Echocardiography

All patients underwent 2-dimensional and Doppler echocardiographic examination using a commercially available ultrasound system (MyLab Class C, Esaote, Genova, Italy). The left ventricular EF was measured using Simpson’s modified biplane method.

Speckle tracking analysis for estimation of global longitudinal peak systolic strain (GLPSS) was performed using dedicated software (MyLab Desk, Esaote).

Brain magnetic resonance imaging

Magnetic resonance imaging (MRI) acquisition was performed on a 3T Siemens Skyra scanner (Siemens, Erlangen, Germany) with a 24-channel head-neck coil.

Segmentation of the T1-weighted 3-dimensional brain scans was performed using the NeuroQuant (version 2.3; CorTechs Labs Inc, San Diego, California, United States), which is a fully automated, FDA-cleared tool.4

Statistical analysis

All analyses were performed with SPSS (version 23.0, IBM Corp, Armonk, New York, United States). Continuous data are reported as a median and interquartile range (IQR). Baseline characteristics are reported as standard descriptive statistics. Differences between groups were estimated using the t test for data with normal distribution and with the Mann-Whitney test for data without normal distribution. The interrelationships between metrics were assessed using the Spearman rank correlation analysis. All analyses were 2-tailed, and a P of less 0.05 was considered significant.

Results

The median age of all participants was 63 years (IQR, 59–66). For group 0, median age was 62 years (IQR, 59–66). This group included 25 women. None of the healthy controls was taking any medication. For group 1, the median age was 64 years (IQR, 59–69), and there were 5 women. Among these patients, reduced left ventricular EF was due to either dilated cardiomyopathy (11 cases) or previous myocardial infarction (14 cases). Patients with HF were receiving the most contemporary pharmacological treatment, that is, an angiotensin-converting enzyme inhibitor or angiotensin-receptor blocker (100% of patients), a β-blocker (92%), an aldosterone inhibitor (84%), a statin (56%), a diuretic (68%). Six patients had diabetes and 10 had hypertension.

Table 1 demonstrates the hemodynamic and morphometric characteristics of the cardiovascular system and brain segments in patients from both groups. There was no difference in age, blood pressure, and body mass index between the groups. Ejection fraction (P <0.0001) and global longitudinal peak systolic strain (P <0.0001) were reduced in patients with HF. Analysis of brain segmentation revealed that only hippocampus volume (6.6 cm3 vs 7.4 cm3, P = 0.01) and putamen volume (10.4 cm3 vs 10.8 cm3, P = 0.03) were reduced in patients with diminished EF. In the whole study population, hippocampus and putamen volume did not differ between women and men (mean [SD], 7.2 [0.9] cm3 vs 7.3 (1.2) cm3, P = 0.6; and 10.4 [1.2] cm3 vs 10.9 [1.5] cm3; P = 0.1, respectively).

Table 1. Critical cardiac, hemodynamic and brain metrics in the studied population
ParameterWhole populationGroup 0Group 1P value

Number of patients

63

38

25

–

Women/men

30/33

25/13

5/20

0.0004

Age, y

62 (59–66)

62 (59–66)

64 (59–69)

0.2

BMI, kg/m2

26.2 (23.8–29.4)

26.4 (23.6–30.7)

25.9 (24.2–28.7)

0.9

BP systolic, mm Hg

136 (115–140)

137 (120–150)

128 (110–145)

0.1

BP diastolic, mmHg

79 (73–89)

80 (74–91)

76 (69–87)

0.09

GLPSS, %

–17.29 (–19.54 to –9.28)

–18.9 (–20.3 to –17.6)

–8.37 (–10.67 to –6.89)

<0.0001

EF, %

60 (35–65)

64 (61–68)

32 (19–38)

<0.0001

TBV, cm3

1165 (1086–1246)

1180 (1034–1197)

1123 (1090–1213)

0.2

Hippocampus, cm3

7.3 (6.6–7.9)

7.4 (6.8–8.2)

6.6 (6.1–7.5)

0.01

Amygdala, cm3

3.2 (2.8–3.4)

3.2 (2.9–3.5)

3.2 (2.8–3.4)

0.4

Caudate, cm3

5.4 (4.8–5.9)

5.4 (4.7–5.9)

5.4 (5.0–5.8)

0.3

Cuneus, cm3

10.3 (9.3–11.3)

10.5 (9.9–11.4)

9.7 (8.9–11.4)

0.08

Cingulate, cm3

14.4 (13.3–15.4)

14.5 (13.4–15.7)

14.3 (13.1–15.3)

0.3

Putamen, cm3

10.4 (9.8–11.4)

10.8 (10.1–11.8)

10.4 (9.3–11.3)

0.02

Thalamus, cm3

15.0 (14.1–16.3)

15.0 (13.9–16.7)

15.0 (14.2–16.1)

0.7

Data are presented as a median and interquartile range.

Abbreviations: BMI, body mass index; BP, blood pressure; EF, ejection fraction; GLPSS, global longitudinal peak systolic strain; TBV, total brain volume

Analysis of bivariate correlation between volume of segmented brain structures and GLPSS and EF in the mixed population of patients with normal and diminished left ventricular function showed that only areas affected by volume loss demonstrated a correlation with markers of LV function. Hippocampus volume correlated inversly with GLPSS (r = –0.3, P = 0.04) and almost significantly and positively with EF (r = 0.24, P = 0.058). Putamen volume was associated with both GLPSS and EF (r = –0.3, P = 0.03; r = 0.3, P = 0.008, respectively).

Discussion

We demonstrated that patients with systolic HF present brain volume loss. Not all gray matter area associated with cognitive function showed morphological changes in comparison with controls. Interestingly, in a mixed population of healthy controls and patients with HF, markers of left ventricular function were associated with brain segments with volume loss but not with those unaffected.

It is well established that patients with HF are more frequently prone to depression and cognitive impairment. Moreover, both functional and organic brain abnormalities are found more often in these patients than in healthy controls. Almeida et al5 showed that adults with HF had impaired immediate and long-term memory in comparison with controls. Moreover, HF is associated with other changes, namely lower volumes of brain regions that are important for cognitive and emotional processing. Putamen plays a significant role in emotional processing, memory function, and motor planning. Kumar et al6 demonstrated in 15 patients (mean age, 54 years; low EF mean, 27%) a global and regional putamen loss in comparison with controls. It was shown that cingulate cortex plays an essential role in mediating cognitive influences on emotion. Almeida et al5 presented evidence of gray matter loss in the cingulate region in 35 patients with HF. Moreover, these patients demonstrated impairment of cognitive skills in comparison with healthy controls. Hippocampus is a brain structure deeply involved in memory and mood. Woo et al2 recently demonstrated in 17 patients with HF and low EF (median, 28%) a significant hippocampal volumetric loss in comparison with controls. Several reports demonstrated other types of injury in regions involved in cognitive function (caudate, cuneus, or amygdala). Regional cerebral blood flow was diminished in the cuneus region in patients with HF.7 The amygdalar region is involved in emotional stress, and its activity was associated with an increased risk of cardiovascular complications.8 Caudate nuclei have memory, learning, and cognitive functions. Woo et al9 demonstrated in 17 patients with HF (mean age, 54 years; mean EF, 28%) decreased functional connection (interaction ability via white matter connections) between brain sites in comparison with healthy controls.

In our study, brain segmentation was performed with the volumetric assessment of gray matter regions involved in cognitive, memory, learning, and motor function. We did not observe a difference in total brain volume between patients with severe impairment of left ventricular systolic function and healthy controls. Interestingly, out of 7 evaluated regions, only hippocampus and putamen volumes were reduced in patients with HF. It was also observed that regions with volume loss were correlated with LV systolic strain and EF. Of note, similarly to other reports, GLPSS was more sensitive in analyzing such an association.10

In a similar investigation into the anatomic distortion of brain structure in patients with HF, Vogels et al11 reported more white matter hyperintensities, lacunar infarcts, and temporal lobe atrophy in comparison with controls. One reason for brain damage observed in these individuals could be the impairment of blood flow because, despite the fact that cerebral blood flow (CBF) is maintained across a wide range of distending pressure, the autoregulatory mechanisms of CBF are blunted in patients with HF. Moreover, heart transplantation improves CBF in HF patients. These and other observations sparked recent interest in the association between heart function and brain structure in a more diverse population. Russo et al10 demonstrated that lower global longitudinal LV strain, but not EF, was independently correlated with subclinical brain disease in a community-based cohort without overt heart disease.

There are several other possible explanations on the mechanisms responsible for volume loss of brain regions in patients with HF including hypoxia, cerebral microinfarctions, impairment in cerebrovascular regulation, comorbidities (diabetes, hypertension), sleep apnea, or chronic inflammation.

The present study does not prove a causal link between cardiovascular function and brain architecture; however, it is possible that GLPSS can serve as a marker of cerebral risk burden similar to that provided by EF assessment.

Study limitations

This study has several limitations. First, we selected a mixed population of patients with contrasting EF; however, it is impossible to find healthy controls with low EF. Second, we estimated only a limited number of brain structures. Third, we did not evaluate white matter lesions and other markers of subclinical brain injury. Fourth, we did not balance study groups for gender differences. However, recent evidence did not support the presence of sex-related dimorphism in brain substructure volume.12

In summary, we found that only particular gray matter brain structure showed diminished volume in patients with HF. Metrics of LV function such as global longitudinal peak systolic strain and EF are correlated with cortical gray matter volume of brain segments related to volume loss, but not with those unaffected. These findings need to be confirmed in a larger study population.