Heart rate variability as a biomarker in chronic chagas cardiomyopathy patients with or without concomitant digestive involvement, for prediction of rassi score risk classes

Introduction

The pathogenesis of Chronic Chagas Cardiomyopathy (CCC) is not yet fully elucidated. However, dysautonomia is one of the factors involved, in addition to being the essential mechanism in the pathogenesis of the Digestive Form of Chagas Disease (DFCD). The prognostic value of dysautonomia remains speculative, and there are no correlative studies of dysautonomia in CCC and DFCD.

Purpose

This study has three aims: a) to investigate in patients with CCC the relationship between cardiac dysautonomia, indirectly studied by heart rate variability (HRV), and the prognostic stratification assessed by the Rassi score; b) to compare the HRV in groups with isolated CCC and with the mixed form, i.e. CCC associated with DFCD; c) to evaluate the power of combining HRV indices to predict the risk class of each patient, using machine learning.

Methods

Thirty-one patients with CCC were classified into three risk groups (low, intermediate and high) according to their Rassi score and had two electrocardiograms (ECG) recorded, i.e. the conventional 12-lead and a single lead, the latter for a period of 10 to 20 minutes. From the single lead ECG, two equally sized RR series were generated and 31 HRV indices were calculated from each. The HRV was then compared between the three risk groups and also regarding the presence or not of concomitant digestive impairment. Taking HRV indices as inputs, four machine learning models were compared in its ability to predict the risk class of each patient. A previous step of attribute selection (sequential feature selection) was applied to identify the most relevant HRV indices for each algorithm.

Results

Comparing the HRV indices in the three risk groups obtained with the Rassi score, the phase entropy is decreased [0.91 (0.90, 0.91) vs 0.87 (0.86, 0.89); p=0.039] and the percentage of inflection points is increased [66.4 (63.5, 71.2) vs 58.2 (53.4, 63.3); p=0.032] in patients in the high-risk group, compared to the low-risk group. Of the 31 patients with CCC, 14 had the mixed form of the disease, i.e. with associated digestive impairment. In the latter, the triangular interpolation of the RR interval histogram decreased significantly [78.1 (62.5, 101.6) vs 121.1 (80.1, 146.5), p=0.046], while the absolute power in the low-frequency band decreased with strong trend to statistical significance [28.5 (17.1, 97.5) vs 86.9 (44.1, 171.7), p=0.06]. The best predictive model for each risk group was obtained with the Support Vector Machine, reaching an overall F1-score of 0.61.

Conclusions

The worst prognosis, indicated by the Rassi score, is associated with increased heart rate fragmentation. The combination of HRV indices enhanced the accuracy of the risk stratification. Compared to CCC the mixed form of Chagas' disease displays a decrease in the components of slow heart rate oscillation, suggesting a higher degree of sympathetic autonomic denervation associated with parasympathetic impairment.

Funding Acknowledgement

Type of funding sources: Foundation. Main funding source(s): São Paulo Research Foundation (FAPESP)

P187 Multimodality in tuberculous constrictive pericarditis

CASE PRESENTATION

K.C.P., a 26 y.o. female, presenting dizziness and progressive dyspnea since 9 months ago. Physical examination showed hepatomegaly at 2 centimeters below the right inferior costal border, but without edema, cardiac murmurs or other findings. Electrocardiogram showed atria overload and diffuse ventricular repolarization abnormality. Chest X-ray revealed normal sized cardiac silhouette but with signs of pericardial calcification. Transthoracic echocardiogram revealed: enlargement of both atria, no signs of myocardial left ventricular (LV) hypertrophy; dilated inferior vena cava with minimal respiratory variation; septal bounce; septal e´= 17.20 cm/s, lateral e´= 6.09 cm/s; E/e" septal ratio = 3.9; E deceleration time = 144 ms; thickening and hyper-refringence of the pericardium with calcification adjacent to the lateral and inferior walls of the LV and the free wall of the right ventricle. For evaluation of thickness and extent of pericardial involvement, computed tomography (CT) was performed, showing gross calcifications of the pericardium, mainly in basal and lower portions, without pericardial effusion. Cardiac magnetic resonance (CMR) imaging with late gadolinium enhancement revealed areas suggestive of active inflammation adjacent to the basal wall of the LV. With this evidence of active inflammatory activity, the patient was treated empirically against the etiologic agent for tuberculosis. Because of progressively severe edema and dyspnea pericardiectomy was warranted providing relief of symptoms. DISCUSSION: The echocardiogram is the initial image exam for diagnosis and monitoring of pericardial conditions. It is a widely available, low-cost method that does not use ionizing radiation and allows a complete morphological and functional evaluation of the heart. However, in up to 20% of cases, pericardial thickening may not be detectable at echocardiography. CT allows a more accurate assessment of pericardial thickening, while CMR allows detection of active inflammatory process. CONCLUSION: A typical and illustrative clinical case of constrictive pericarditis is presented, where the multimodality of cardiac imaging was decisive for the diagnostic and therapeutic delineation.

P1284 The role of three-dimensional echocardiography in the evaluation of geometry and function of the left ventricle in patients with chronic chagas cardiomyopathy

BACKGROUND

A hallmark of chronic Chagas" cardiomyopathy (CCC) is the early appearance of regional wall motion abnormalities of the left ventricle (LV), especially in the lateral and apical regions. The morphological and functional assessment of LV by two-dimensional (2D) echocardiography, besides depending on geometric assumptions, may not include the most affected segments. The three-dimensional (3D) method offers advantage for the quantification of chamber volumes, but its role in the assessment of patients with CCC is not well established. PURPOSE: The aim of this study is to evaluate the concordance between 3D and 2D methods in the quantification of left ventricular end-diastolic volume (LVEDV) and left ventricular ejection fraction (LVEF) in patients with CCC. METHODS: 44 patients with CCC were selected. Seven subjects were excluded from the analysis (3 for atrial fibrillation, 1 for frequent ventricular extrasystoles, 3 for inadequate acoustic window), resulting in 37 patients. Subjects were submitted to 2D and 3D transthoracic echocardiography by a single experienced examiner following the recommendations from the chamber quantification guidelines. The concordance between the methods was evaluated using the Lin concordance correlation coefficient (CCI) and the Bland-Altman graph analysis. Descriptive analyses of the percentage of individuals reclassified as to altered or to non altered-LVEDV and LVEF values were performed using normal values of 3D as a reference. RESULTS: Age = 58 ± 12 years, 54% men. LVEF-2D = 47 ± 9% and LVEDV- 2D = 73 ± 18 ml / m². The CCI for the measurements of LVEDV and FEVE by both methods was 0.861 and 0.833, respectively. The mean difference found was 3 mL for LVEDV and 3% for LVEF. The 95% agreement limits found were (-14, 20 ml) for LVEDV and (-5; 11%) for LVEF. When assessed with 2D, 27% of individuals previously classified as having LV enlargement were reclassified as having normal LV volumes using the 3D method. Conversely, regarding LVEF, 16% of individuals considered normal by 2D were reclassified as having reduced LVEF with the 3D method. CONCLUSION: In individuals with CCC measurements of LVEDV and LVEF by 3D and 2D methods show high agreement. However, 3D evaluation allows the reclassification of 27% and 16% of the individuals regarding the presence of LV dilatation and LV systolic dysfunction, respectively.

Abstract P1284 Figure. Bland-Altman plot - LVEF

Alcohol distribution in different postmortem body fluids

Aims: The determination of postmortem ethanol is one of the most frequently requested analyses in forensic toxicology and of extreme importance, especially when the concentration is found to be above the legal level for intoxication at one sampling site and below that level at another sampling site. Because of the unavailability of blood samples for toxicological analysis or even contaminated samples, there is an enormous effort to find alternative sampling sites, such as vitreous humor for ethanol analysis. The main purpose of this study was to establish correlations between urine and blood alcohol concentrations collected from different sites and vitreous humor. Methods: Ethanol concentrations were determined in specimens of heart, subclavian and femoral blood, urine and vitreous humor, collected from 21 cadavers who were victims of different causes of death. Determinations of ethanol were performed in duplicate using capillary gas chromatography/flame ionization detector and headspace techniques. Results: Statistical analysis of the results indicated that there were no significant differences among urine and blood samples collected from different sites compared to vitreous humor. Regarding vitreous humor ethanol concentration, Pearson's correlation coefficient was 0.97 for femoral blood and urine, 0.96 for heart blood and 0.94 for subclavian blood. The results demonstrated that all the fluids tested against vitreous humor significantly correlated with P (associated probability for the used correlation tests) B<0.05. Conclusions: Vitreous humor can be considered as an alternative sample to urine and blood specimens.