Keywords
INTRODUCTION
Chagas disease is one of the most worrying health care problems in Latin America and is the main cause of infectious myocarditis1; 18 to 20 million people are infected in Latin America.2 Chagas disease is characterized by 3 phases--acute, indeterminate, and chronic3; the chronic phase tends to display a slow course in which chronic chagasic cardiomyopathy is the principal manifestation. The clinical presentation in the phase of chronic myocarditis is diverse, and heart failure is the most advanced, since it is indicative of extensive, severe, and irreversible damage to the myocardium preceding death.4 Asymptomatic left ventricular dysfunction (ALVD) is an earlier stage, and the development of heart failure is indicative of worsening not only of the patient's symptoms but also of prognosis for survival.5 Five-year mortality is approximately 70% in patients with Chagas disease and heart failure.6 Development of the syndrome is linked to compensatory mechanisms triggered during the early stages of damage to the myocardium and that are initially adaptive before subsequently becoming harmful.
There are few methods available with which to assess ALVD in chronic chagasic cardiomyopathy and predict progression to heart failure, and those that are available are incomplete. We therefore sought to establish predictors of progression towards heart failure using simple clinical and echocardiographic variables normally used in follow-up of these patients. As a secondary objective, we analyzed overall mortality in this patient group.
METHODS
A prospective study was undertaken in our regional referral center for Chagas disease between 1993 and 2006 in which 95 patients aged over 18 years with a diagnosis of chagasic cardiomyopathy and ALVD were consecutively enrolled for subsequent follow-up. The study group was selected from a population of 2990 patients with ALVD (New York Heart Association [NYHA] class I) and serologic diagnosis of Chagas disease seen in our hospital and comprised 95 patients (3.18%). Stress tests in those patients indicated a functional class equivalent to NYHA class I, with a maximum load achieved of at least 6 metabolic equivalents (MET) and its equivalent in kg; 27 (0.9%) had heart failure (NYHA II/IV); 762 (25.5%) only had electrical conduction abnormalities or arrhythmias; and 2033 (68%) were in the indeterminate phase of the disease.
Patients were excluded from the study if they had concomitant heart disease, such as congenital, hypertensive, ischemic, or alcoholic heart disease, valve disease, and others.
Diagnosis of the disease was established with a least 2 out of 3 positive serologic tests for Trypanosoma cruzi7: enzyme-linked immunosorbent assay, indirect hemagglutination, and immunofluorescence test. Deterioration of left ventricular systolic function was assessed by echocardiography according to the measures and variables employed in our hospital and also taking into account the subjective assessment of the operator regarding general systolic dysfunction as mild, moderate, or severe.
The following were performed at the beginning of follow-up: serology for Chagas disease, baseline electrocardiogram (ECG), frontal chest x-ray, stress test, and M-mode, and 2-dimensional echocardiography. During follow-up, patients attended appointments every 2 months or in the event of symptoms. ECG was repeated at each appointment and other tests were done once a year.
The following variables were considered in the study: sex, age on enrollment, length of follow-up, appearance or development of heart failure, ECG abnormalities associated with Chagas disease, complex ventricular arrhythmias (ventricular pairs, sustained, and nonsustained ventricular tachycardia), diastolic and systolic left-ventricular diameter, presence of segmental lesions in the left ventricle, shortening fraction, ejection fraction, E-point septal separation, and left ventricular systolic dysfunction defined semiquantitatively by the cardiologist responsible for echocardiography as mild, moderate, or severe.
The following were considered ECG anomalies related to Chagas disease: complete right bundle branch block, left anterior hemiblock, complete left bundle branch block, ventricular pairs, sustained and nonsustained ventricular tachycardia, areas of electrical inactivation, atrial fibrillation or flutter, atrial tachycardia, sinus bradycardia (<50 beats/min), type 2 second degree and third degree atrioventricular block, and definitive pacemaker implantation; all of these abnormalities displayed an incidence that differed significantly from the ECG findings in healthy individuals.8
Diagnosis of left ventricular systolic dysfunction was established when the cardiologist responsible for echocardiography reported mild, moderate, or severe dysfunction (defined semiquantitatively) accompanied by at least 1 of the following quantitative parameters: E-point septal separation >10 mm, shortening fraction <25%, and ejection fraction <45%. The reference values used in our hospital for normal left ventricular systolic function are an E-point septal separation <7 mm, shortening fraction >28%, and ejection fraction >55%.
Heart failure was defined according to Framingham criteria, establishing the simultaneous presence of at least 2 major criteria or 1 major and 2 minor criteria.9
The study protocol was approved by the institutional review board and patients consented to inclusion in the study.
Statistical Analysis
Continuous variables were expressed as means (SD) or median (interquartile range) according to whether they exhibited a gaussian distribution in the Shapiro-Wilk test, and categorical variables were expressed as proportions and/or percentages. The primary outcome measure was the development of heart failure and the secondary measure was overall mortality. Univariate and multivariate Cox regression were used to establish predictors of the development of heart failure. The assumption of proportionality was assessed by Kaplan-Meier analysis. The variables that displayed a significant association with the dependent variable (P<.10) in the univariate analysis were incorporated in the multivariate model to confirm the independent predictive value of each variable, considering values of P<.05 to be statistically significant. The hazard ratio (HR) and 95% confidence interval (CI) was calculated for each variable included in the Cox regression model. The area below the receiver operating characteristic (ROC) curve was used to determine the sensitivity and specificity of the continuous variables that displayed statistical significance and the curve of the risk function (derived from the Kaplan-Meier curve) was used to display the differences. Analysis was performed with the programs Statistix 7.0 Analytical Software and SPSS version 10.0 Statistical Analysis Software for Windows (SPSS Inc, Chicago, Illinois, USA).
RESULTS
The follow-up period for the 95 patients with chronic chagasic cardiomyopathy and ALVD had a median length of 63 months and an interquartile range of 32 to 110 months. The mean age of the patients was 55 (11) years and the group included 58 (61%) men and 37 (39%) women. The baseline characteristics of the patients included in the study are shown in Table 1. Severe systolic dysfunction was observed in 17% of the population and in 20% when considering an E-point septal separation of at least 20 mm as an indicator of severe systolic dysfunction. Most of the patients had mild or moderate systolic dysfunction on inclusion in the study. In 96.8% of cases, the patients received treatment with angiotensin converting enzyme inhibitors or angiotensin II receptor antagonists.
Twenty-seven patients (28%) developed heart failure during follow-up. Table 2 shows the results of the univariate analysis. The variables that were not associated with the primary outcome measure (development of heart failure) were sex, ECG abnormalities, systolic left ventricular diameter, diastolic left ventricular diameter, presence of segmental lesions, complex ventricular arrhythmias, shortening fraction, ejection fraction, and moderate depression of systolic function. Age at enrollment, mild or severe depression of left ventricular systolic function, and E-point septal separation displayed statistically significant differences.
Table 3 shows the results of the multivariate analysis (Cox regression). Age and mild depression of left ventricular systolic function lost their independent predictive value for heart failure, while the presence of severe left ventricular systolic dysfunction and greater E-point septal separation maintained their prognostic value. The sensitivity and specificity of those variables was calculated: severe left ventricular systolic dysfunction had a sensitivity of 22% and a specificity of 85%. We used a ROC curve to determine the best cutoff for the E-point septal separation and selected a value of 17 mm, which was associated with a sensitivity of 63% and a specificity of 75% for prediction of the development of heart failure (Figure 1). The area below the ROC curve was 0.68 (a test with no discriminative power has a value of 0.50). Figure 2 shows the risk of developing heart failure in relation to severe deterioration of systolic function.
Figure 1. Receiver operating characteristic curve for the variable E-point septal separation in relation to the development of heart failure in patients with chronic chagasic cardiomyopathy and asymptomatic left ventricular dysfunction. An E-point septal separation of 17 mm represents the cutoff with the best combination of sensitivity and specificity.
Figure 2. Risk function for the development of heart failure in patients with chronic chagasic cardiomyopathy and asymptomatic left ventricular dysfunction in relation to severe deterioration of systolic function.
The mortality at 63 months of follow-up was 3/95 (3%) patients who continued to have ALVD and 10/27 (37%) patients who had developed heart failure. The time from inclusion in follow-up to development of heart failure was 55 months, with an interquartile range of 23 to 117 months. The mean follow-up from the appearance of heart failure to death was 46 (37) months.
DISCUSSION
Our study assessed the possible predictors of heart failure in a sample of 95 patients with chronic chagasic cardiomyopathy and asymptomatic ventricular dysfunction. Patients with severe left ventricular dysfunction had a 2.5-fold greater risk of developing heart failure than those with mild or moderate deterioration of left ventricular systolic function. In studies designed to assess mortality, this was found to increase with deterioration of left ventricular systolic function.10 An E-point septal separation of at least 17 mm was associated with a 3.5-fold increased risk of developing heart failure compared to those in whom the distance was shorter. The data obtained in our hospital based on 2990 patients with Chagas disease reveal a wide range of clinical presentations of the disease and a small number of patients who progress to more severe forms of the heart disease.
Various echocardiographic studies have demonstrated that segmental abnormalities of left ventricular wall motion in chronic chagasic heart disease contribute to general deterioration of left-ventricular systolic dysfunction,11 which in turn contributes to poor prognosis in those patients.12-14 The number of affected segments generally shows a good correlation with general deterioration of systolic function.15 It is known that the most common site of segmental abnormalities in chronic chagasic cardiomyopathy is the apex of the left ventricle.11-16 This leads to shortening of basal segments being conserved or little affected, due to the compensatory mechanisms that are triggered. Depending on the method used to assess ventricular function--by measurement of systolic and/or diastolic left ventricular wall diameter, shortening fraction, and ejection fraction (Teichholz)--deterioration may be underestimated or overestimated, since those measurements are made in the basal segments and do not offer an adequate assessment of general function in ventricles with regional differences in contraction (Chagas disease and other segmental heart diseases).17 In our study, these variables did not display statistical significance as predictors of heart failure, a finding that may be explained by the regional differences in contraction and the homogeneity of those variables in our sample of patients with ALVD. The E-point septal separation and severe general systolic dysfunction, as defined by the cardiologist, were the only factors with prognostic value for heart failure in the multivariate analysis. The E-point septal separation has the advantage of being a measure of general systolic function that is relatively independent of regional abnormalities in left ventricular wall motion,18 with a good correlation with angiographic ejection fraction.19,20 In addition, it displayed acceptable sensitivity and specificity when analyzed using a ROC curve. The semiquantitative assessment by the operator (cardiologist trained in echocardiography) can offer a more general interpretation of ventricular function, since it considers both M-mode measurements at the base of the left ventricle (shortening fraction, systolic diameter of the left ventricle, etc) and assessment of general and regional motion, influenced by the various echographic views.21 This may explain the good correlation with ejection fraction estimated from radionuclide studies22 and may even be better than that calculated by the Simpson method in terms of correlation with estimates from radionuclide imaging.23 Calculation of the ejection fraction using the Simpson method requires adequate visualization of the endocardial borders and a considerable amount of time, which may not always be available. Due to technical limitations, we do not use this method, and the ejection fraction was calculated according to the Teichholz formula, in the knowledge of is limitations in the presence of segmental lesions. Another limitation was the lack of cardiac Doppler studies, which can improve the assessment of left ventricular systolic and diastolic function.
None of the clinical variables such as age and ECG abnormalities associated with Chagas disease displayed prognostic value for the development of heart failure in our study. In general, age is associated with greater mortality in patients with heart failure, but this was not the case in our group of patients with ALVD. ECG findings also showed no statistical significance for the prediction of heart failure, in contrast to the prognostic value associated with cardiac death.24 Likewise, complex ventricular arrhythmias were not associated with development of heart failure, even though they are known to be closely associated with deterioration of left ventricular function and increased risk of death.13,25 Segmental lesions did not seem to be independent predictors of heart failure, at least not in the patient sample studied here, with no classification of the lesion according to type or location.
Selection of patients with ALVD appears to have led to homogeneity of many of the variables studied, and they did not display an association with the outcome measures of the study. These same variables may have had a different significance if the outcome measure was mortality and the study group had heart failure.
CONCLUSIONS
In patients with chronic chagasic cardiomyopathy and ALVD, 2 easily obtained echocardiographic variables--E-point septal separation and severe left ventricular dysfunction defined by the operator--can predict the development of heart failure.
See editorial on pages 105-7
ABBREVIATIONS
ALVD: asymptomatic left ventricular dysfunction
ECG: electrocardiogram
FC: functional class
LV: left ventricle
NYHA: New York Heart Association
ROC: receiver operating characteristic
Correspondence: Dr. M.A. Petti.
Pje. General Hornos 2771. (ZC 1640) Martínez. Buenos Aires. Argentina.
E-mail: mapetti@intraed.net
Manuscript received May 18, 2007.
Accepted for publication November 2, 2007.