Longitudinal Study of Patients With Chronic Chagas Cardiomyopathy in Brazil (SaMi_Trop Project)

RESEARCH STRATEGY

SIGNIFICANCE

The Sao Paulo-Minas Gerais Tropical Medicine Research Center (SaMi-Trop) consists of a network of collaborating scientists in the States of Minas Gerais and São Paulo which has been established for the purpose of developing and conducting research projects on neglected infectious diseases in Brazil, with an initial focus on Chagas Disease biomarker, discovery and validation. The SaMi-Trop represents a major challenge for focused research in neglected diseases, with knowledge that can be applied in primary health care.

Chronic Chagas Cardiomyopathy (CCC) is the most important clinical presentation of Chagas Disease (ChD), eventually affecting approximately 20% to 30% of those in the chronic phase of the disease. CCC comprises a wide range of manifestations, including heart failure, arrhythmias, heart-blocks, sudden death, thromboembolism and stroke [1]. Clinical presentation typically varies widely according to the degree of myocardial damage and most patients present a mild form of heart disease, frequently characterized only by the presence of asymptomatic abnormalities on the ECG or in other complimentary exams [2]. Nonetheless, when heart failure and/or severe arrhythmias manifest, the prognosis is ominous, with high and premature mortality rates, typically in adult male patients,1 but also in the elderly [3]. Indeed, when compared to patients with idiopathic cardiomyopathy, patients with CCC have poorer survival, irrespective of other clinical and echocardiographic parameters [4].

Thus, the clinical course of Chagas heart disease is variable and identifying patients who are at risk of progressive CCC and dying remains a challenge. Several longitudinal studies performed in patients in the chronic phase of ChD demonstrated that many individual characteristics predict an unfavorable prognosis. This subject was systematically reviewed a few years ago and concluded that most studies were described in insufficient detail and/or lacked minimal standards in reporting results. Results have been somewhat inconsistent in different studies, and the best combination of clinical and laboratory variables predictive of mortality has not yet been established [5].

A potentially helpful score to predict death was developed to permit the stratification of risk in Chagas heart disease [6]. Six independent prognostic factors were identified, and each was assigned a number of points: New York Heart Association class III or IV (5 points), evidence of cardiomegaly on radiography (5 points), left ventricular systolic dysfunction on echocardiography (3 points), non-sustained ventricular tachycardia on 24-hour Holter monitoring or stress testing (3 points), low QRS voltage on electrocardiography (2 points), and male sex (2 points) [6. Patients were classified in three risk groups according to the final score: low risk (0 to 6 points), intermediate risk (7 to 11 points), and high risk (12 to 20 points). In the original study, the 5-year mortality rates for these three groups were 2 percent, 18 percent, and 63 percent, respectively.6 Although the score was successfully validated in independent cohorts, [6,7] some important issues may arise. The derivation of the score is complex and it is not easy to calculate during the medical consultation. To be calculated, the patient must be submitted to several exams and procedures, including clinical examination, ECG, Chest x-ray, echocardiogram and Holter monitoring or stress testing. Most of these tests are not available in the rural areas, where most of ChD patients live.

A simple score, based on widely available and low-cost tests, which could stratify the risk of death without complex calculations, might be useful in the primary care setting, where echocardiograms and Holter monitoring are not widely available. Easily obtained variables, such as clinical features, BNP levels and ECG measurements, are good candidates as biomarkers for developing a new simple predictive rule. Moreover, new biomarkers in blood, related mainly to inflammation and the immune response, such as TNF-alpha, adhesion molecules and other cytokines, may be determined to be potent prognostic factors and may be useful in the future for further improvement in risk stratification in CCC.

METHOD

Design

Non-concurrent prospective open cohort study, with minimal follow-up of two years (baseline and 24 months).

Outcome

Primary: death Secondary: hospitalization due to cardiac conditions; new ECG abnormalities

Study Population

Participants will be selected from patients evaluated by the Tele Minas Saude (TMS), the state Tele Health System. Since the TeleHealth System is currently acting in support of the Primary Care system, these patients are under the care of primary care physicians of the Health Family Strategy, a public primary care program that has a high coverage in the state of Minas Gerais.

The inclusion criteria are to be a client of the TMS TeleHealth System, be >18 years old, have a positive history for T cruzi infection or Chagas disease, have an abnormal ECG and to consent to participating in the research program.

The exclusion criteria include pregnancy or breast feeding, and any severe condition with ominous prognosis that indicates a life expectancy of less than two years.

Sample size for patient accrual

Considering the minimal number of events per variable acceptable in a proportional hazards regression analysis of 10 events per variable,8 and maximal number of studied variables of ten, the number of events would be 100 in the whole study. Since the prediction model has to be developed and validated and the whole sample will be divided in two, the number of events should be 200. For a 2-year follow-up period and annual mortality rate of 5% in CCC (10% in 2 years), the calculated sample size is 2,000 subjects.

Since the TMS Telehealth system performs 25,000 ECGs each month (300,000 a year), that ECG abnormalities typical of CCC are observed in 5% of these exams (1,250), and projecting that 10% of these patients would be recruited and agree to participate in this research program, the calculated sample size of 2,000 patients will be recruited in a 16 months' accrual period.

Follow-up

Follow-up will be conducted on a yearly basis, with standardized interviews and new ECG exams performed after two years of baseline. Vital status and hospitalization will also be ascertained by linkage of the study database with mortality and hospitalization databases of the Ministry of Health.

Recruitment, initial evaluation and measurements

Patients will be recruited from among those that regularly attend the public primary care service of State of Minas Gerais in places served by TMS Telehealth System. This system provides teleconsultation and Electrocardiogram services (TeleECG), including interpretation of all ECGs by a cardiologist in the University Hospital. A standardized simple form is completed before the transmission of the ECG data to the University hospital and the question "Do you know if you have Chagas disease?" will be added to this questionnaire. Those who answer yes to this question and have abnormal ECG will be selected. A second visit of the selected patients will be scheduled for performing a short questionnaire (including the evaluation of the functional class) and for collecting blood samples. Blood samples will be sent to a central laboratory for T. cruzi serology and BNP measurements, which will be done by standard techniques. Other blood samples, including plasma for protein and antibody studies and stabilized whole blood lysates for T cruzi PCR (Polymerase Chain Reaction), DNA, will be stored at -80o C to enable batched testing for other biomarkers.

Electrocardiogram services (TeleECG)

The ECG performed in the town of origin of the patients will be sent to be assessed by the TMS Telehealth System specialists. The Telecardiology activities were developed using a digital electrocardiograph installed in the towns, at the Basic Health Units of the public healthcare system. The examination was always carried out after being requested by a physician. The ECG will be sent using communication software developed by the TMS Telehealth System and will be analyzed by a on-duty cardiologist. The report will be sent to the Primary Care facility and the ECG tracings will be transferred to ECG reading center. In this ECG Reading Center, a software developed by Prof. Peter Mac Farlane, from Glasgow University, will analyzes and codify the ECG (by the Minnesota Code) and compare to the previous ECG, according to a validated algorithm [9]. All abnormal codes will be reviewed by an experienced cardiologist.

Data analysis

The standardized mortality ratio (SMR) analysis and a survival analysis will be conducted. The outcome death is death. Also, a survival analysis considering the secondary outcomes - hospitalization due to cardiac conditions and new ECG abnormalities - will be run. The development and validation of the prognostic model will follow current recommendations, including the division of the sample in derivation and validation sample, and the use of modern methods to evaluate the performance of the model [10,11]. Additionally, descriptive spatial analysis using geographic distribution for the SMR and a Bayesian spatial model for small area data will be run.

Innovation

Beyond the relevance of the main goal of this study, it has many unique and innovative aspects, related to the use of knowledge and infra-structure developed for other projects conducted by our research group:

1. The possibility of testing new biomarkers detected in other branches of this study or in the NIH-sponsored REDS-II/III Chagas disease study.
2. The use of the infrastructure and the logistics of the TMS program, a state-wide TeleHealth System coordinated by the University Hospital in the Brazilian state of Minas Gerais
3. The use of the methodology of linkage between our databases and administrative and mortality databases from the Health Ministry to complement the follow-up of the selected patients
4. A standardized codification system of the ECGs, using the facilities of the ECG Reading Center of the long-term longitudinal study on the health of 15,000 adult public servants from Brazilian Universities (ELSA-Brazil).

New biomarkers ChD has a complex physiopathology and many immunological, inflammatory, autonomic and micro-vascular processes have a role in the development or the progression of the disease [12]. The investigators have described increased levels of macrophage inflammatory protein (MIP-1alpha) and Tumor necrosis factor-alpha (TNF-alpha) in ChD patients, [13-15] and that those patients with more severe forms have more elevated levels [13]. Recently, in the ChD study (REDS-II), were studied 600 ChD patients and 500 controls searching for biomarkers related to the disease and CCC. In a preliminary analysis, several biomarkers including high sensitivity troponin, myoglobin, soluble vascular cell adhesion molecule 1 (sVCAM-1), soluble intracellular cell adhesion molecule (sICAM) and myeloperoxidase showed abnormal patterns in T cruzi seropositive and parasitemic donors and CCC patients. These and other biomarkers, including those identified in the transcriptome analysis are candidates to be studied in the whole cohort of 2,000 CCC patients in TMS program or, alternatively, using nested case-control design.

Tele Minas Saude (TMS) Program, a state-wide public TeleHealth System

The TMS program is a partnership between the Minas Gerais State Health Department and the State Research Agency (FAPEMIG) with six public universities of Minas Gerais: Universidade Federal de Minas Gerais (UFMG), Universidade Federal do Triângulo Mineiro (UFTM), Universidade Estadual de Montes Claros (UNIMONTES), Universidade Federal de Juiz de Fora (UFJF), Universidade Federal de São João del-Rei (UFSJ) and Universidade Federal de Uberlândia (UFU), forming the Teleassistance Network of Minas Gerais. Created from a telehealth pilot project in cardiology (2006), later expanded to other specialties and villages and transformed into permanent telehealth service, subsidized by the state government in 2009 [16-18].

The objective is to support infrastructure projects of state public health, initially deployed to support the Primary Health Care and Family Health Teams [16]. At that stage, had as main objective to assist professionals in their daily work to improve access to specialized services, reducing and qualifying referrals of patients to secondary and tertiary care. The main activities offered are a) duty of teleconsultation offline in various medical specialties, nursing, dentistry, physiotherapy, psychology, pharmacy and nutrition, b) duty in telecardiology to analyze electrocardiograms (ECGs) and provide teleconsultations online for critical clinical cases c) technical support online and offline, d) permanent education through workshops and training and e) continuous monitoring of system operation in the municipalities.

So far, the program was implemented in 608 municipalities in the state corresponding to 732 telehealth sites. The primary care units in the villages were equipped with computers, electrocardiograph, digital printer and digital camera. Universities were equipped with workstations and videoconferencing equipment. Since 2006, over 20,000 teleconsultations were conducted and 600,000 ECGs analyzed by experts. These telehealth activities prevented the displacement of patients in 78% of the cases [16].

Linkage of databases The Brazilian Health System is characterized by coexistence of a universal public health system (70-80% of coverage) and a supplementary private network. Mortality data from all Brazilians and from hospitalizations in the public health system are stored in large databases owned and managed by the Health Ministry. Investigators developed a methodology for searching these databases for specific subjects using probabilistic linkage methods. The probabilistic strategy we developed has high accuracy, with a sensitivity of 90.6% and specificity 100% [19]. This method will be used in a complimentary way to direct follow-up.

ECG Reading Center ECG interpretation is subjected to substantial intra- and inter-observer variability. To overcome this problem, a standardized ECG code was developed 50 years ago 20 and has been widely used in epidemiological studies. The visual classification of the ECG using the Minnesota code, nonetheless, is time consuming. The investigators created a ECG Reading Center for the Elsa study, a 10-year longitudinal cohort study of 15,000 public employees of Brazilian Universities. For this ECG Reading Center, was installed a software developed by Prof. Peter MacFarlane, from Glasgow University, that analyzes, codifies and compares the ECG with previous tracings, according to a validated algorithm [9]. This system will be used in the present project.

Preliminary Studies

Diagnostic and prognostic value of BNP In 2002, investigators described, in a study published by The Lancet, that blood levels of BNP are inversely correlated to the systolic left ventricular (LV) ejection fraction, the most used index of LV function in ChD [21]. In patients with abnormal ECG and/or chest X-Ray, BNP elevation had a positive predictive value of 80% and a negative predictive value of 97% for the detection of patients with depressed left ventricular ejection fraction [21]. This results were confirmed in the following years by several other studies by our research group [22-26]. A diagnostic strategy including ECG and BNP performed better than the classical approach with ECG and chest X-Ray [24]. Moreover, BNP was also a marker of ventricular arrhythmia [26] and diastolic dysfunction [22,27].

More recently, was described that BNP levels were strong predictors of the risk of death [28] and in a cohort of 1,398 indwelling elderly from the city of Bambuí, in the state of Minas Gerais, Brazil [28]. The hazard ratio for death was 1.27 for each unit of log-transformed BNP level (95% confidence interval (CI: 1.11, 1.45) among infected persons, independent of potentially confounding factors. Infected persons with baseline BNP levels in the top quartile had a risk of death twice that of persons in the bottom quartile (hazard ratio = 2.07, 95% CI: 1.29, 3.32). In the same population, was found that BNP alone or in association with atrial fibrillation had prognostic value for stroke mortality in T. cruzi chronically infected older adults [29].

Electrocardiographic measures and prognostic models The ECG is the single most important exam in CCC. Numerous epidemiological studies have shown that patients with a normal ECG have an excellent medium-term survival [30,31]. The greater the number and severity of ECG alterations registered in a same tracing, the more advanced the myocardial damage possibly is, and the worse the prognosis should be [12]. We have shown that the QRS duration is moderately correlated with the LV ejection fraction [32] and that, during the follow-up of these patients, the increase in the duration of the QRS complex and the appearance of new electrocardiographic alterations may help in identifying patients with a significant decrease (of 5% or more) in left ventricle ejection fraction [33].

Based in the well-known prognostic value of LV ejection fraction and the presence of ventricular tachycardia, [5] was proposed an alternative approach to the well established Rassi's score, [34,35] considering only three risk factors: left ventricular ejection fraction < 50%, ventricular tachycardia at either stress testing or Holter monitoring, and QRS > 133 ms at ECG (or filtered QRS > 150 ms at signal-averaged ECG).34,35 Low risk group has zero or one risk factor (5-year mortality: 1%), intermediate risk, two factors (20% mortality), and high risk, all three factors (50% mortality). This simplified prognostic score had an excellent performance in predicting death in a study with 74 months of follow up (c statistic 0.92) and it may be an attractive alternative to the established six-factor score.

More recently, the investigators showed that T-wave amplitude variability (TWV), a sophisticated ECG-derived measure of repolarization variability, is independently related to the risk of death in ChD [36]. They have also studied the prognostic value of several new echocardiographic measures [37-39] which could be of physiopathological and clinical relevance, but are not possible to be used for ambulatory patients in the rural area.

Kaplan-Meier curves demonstrating mortality in Chagas disease patients Classified according to the presence prolonged of QRS > 133 ms at ECG; Classified according to a risk score in which each of the following factors is worth one point: ventricular tachycardia at either stress testing or Holter monitoring, left ventricular ejection fraction < 0.50, QRS > 133 ms at ECG. Low risk group has zero or one risk factors, intermediate risk, two factors, and high risk, all three factors.

Summarizing, although we and others investigators have extensively studied prognostic markers in Chagas disease and CCC, there is no definite simple, low-cost prognostic model available to be used in the primary care setting. Standard ECG and BNP levels are good candidates to be selected in this simplified predictive model.