IV. CONTENIDO TEMÁTICO
6. TRAZADO DEL DIBUJO
6.3 PREPARACIÓN DE LA CONFIGURACIÓN DE TRAZADO
Chronic heart failure (CHF) is an increasingly widespread, costly and deadly dis- ease that is frequently called a twenty-first century epidemic. Despite advances in modern treatment, the mortality rate in CHF patients remains high. Risk stratifi- cation in patients with CHF remains one of the major challenges of contemporary cardiology [53].
The prevalence of CHF in the US population is 2%. Almost 5 million people are af- fected in total, and 30–40% of all patients die within one year of diagnosis [54]and [55]. Patients with chronic heart failure (CHF) often develop breathing abnormalities, in- cluding various forms of oscillatory breathing patterns that are characterized by rises and falls in ventilation [56] and [57]. Periodic breathing (PB) in CHF is attributed to many factors, including: low cardiac output, which increases the time it takes pul-
monary venous blood to reach central and peripheral chemoreceptors; low lung vol- ume; lung congestion; increased chemoreceptor sensitivity; and the narrow difference between eupneic carbon dioxide tension and the apneic hypoventilatory threshold [58]. Periodic breathing patterns can be classified into Cheyne–Stokes respiration (CSR), and ventilation without apnea [59] and [60]. CSR is characterized by repetitive grad- ual increases and subsequent gradual decreases in ventilation, followed by periods of apnea. It occurs in patients with a variety of diseases and conditions. PB in CHF patients has a similar increase and decrease in ventilation, but there are no periods of apnea. Both ventilatory patterns may have the same mechanistic explanations, as PB is a less severe form of CSR [58].
The prevalence of periodic breathing is as high as 70% in CHF patients [61] and is associated with increased mortality [62], especially in CSR patients [63] and [64]. Accurate risk stratification is crucial to establish the prognosis and to appropriately allocate limited resources for advanced but expensive treatments, including heart transplantation [65]. Breathing patterns are also influenced by wakefulness or sleep, posture, physiological activity and mental activity [66]. Physiological parameters for the characterization and detection of different breathing patterns have been suggested in a number of clinical studies [67] and [68].
The most relevant clinical predictors of the outcome of heart failure patients are the New York Heart Association (NYHA) class, the left ventricular ejection fraction, systolic arterial pressure, and the peak V O2 at cardiopulmonary exercise testing.
Other risk factors include low exercise tolerance, arrhythmias, the cardiothoracic ra- tio and disturbances in the autonomic nervous system. These accepted risk indices perform well in the prediction of mortality due to disease progression, but the pre- diction of sudden cardiac death (SCD) appears more problematic. In hospitalized,
symptomatic CHF patients at high risk of all-cause mortality, death is often due to progressive pump failure. However, in ambulatory patients with less severe CHF, symptoms may be less obvious and death is more often sudden in nature. The impor- tance of being able to identify mildly symptomatic CHF patients at risk from SCD is of paramount importance [69].
Normal breathing cycles range from 3 to 5 s (i.e. 0.20–0.33 Hz). PB patterns have cycles of 25 to 100 s (i.e. 0.01–0.04 Hz) [66]. The origin of the PB pattern is still a matter of debate among researchers. Respiratory modulation frequency appears to be essential for understanding periodic and nonperiodic breathing patterns in CHF patients.
Some of the clinical parameters that are used to estimate the severity of CHF have limitations, due to the heterogeneity and complexity of the syndrome. For example, exertional oscillatory ventilation, which is evaluated during exercise, is associated with a more advanced clinical status, cardiac functional impairment and reduced exercise capacity, and may reflect a more severe alteration of the ventilatory control system [65].
CHF has been related to alterations in breathing pattern [70]. For example, sleep- disordered breathing is associated with an accelerated decline in cardiac function and increased morbidity and mortality [71] in up to 50% of patients with CHF [72]. Several studies suggest that central sleep apnea is highly prevalent among patients with CHF, and the treatment of this sleep-disordered breathing by continuous positive airway pressure could be an important nonpharmacological complement to conventional drug therapy [73] and [74]. In patients with mild to severe CHF, the power of cardiovascular oscillations in the very low frequency band has been reported to be considerably increased by the presence of periodic breathing, which may alter prognosis [75].
Evidence that links exercise capacity to outcome in CHF patients has led to the use of peak oxygen consumption, which is derived from maximal cardiopulmonary exercise testing, and exertional oscillatory ventilation in prognostic assessment. However, these two indices have certain limitations, due to the heterogeneity and complexity of the syndrome [76].
Although CSR is a known sleep-related phenomenon, according to Poletti et al. [63], central apnea and hypopnea also occur in a high percentage of CHF patients during the daytime. Among all the risk factors for daytime CSR, the concentration of plasma NT-proBNP is the best independent predictor of breathing abnormalities. Daytime CSR is significantly associated with more severe clinical impairment, left ventricular ejection fraction and functional capacity, and lower levels of resting. The latter finding suggests that there is a common pathogenesis for daytime and noctur- nal CSR. The absence of obstructive apneas in awake patients during the daytime confirms that in awake states there is adequate stimulation of the dilator muscles to maintain upper airway patency.
In a recent study, Brack et al. [77] proposed that CSR during 10% of the daytime is an independent predictor of death, after adjusting for B-type Natriuretic Peptide (BNP), age, and NYHA class. However, the authors only enrolled 60 patients and employed a long-term recording device, without taking into account patient-to-patient variations in postural changes, speech and physical daily activity.
Maestri et al. [78] studied all the main families of nonlinear methods and found that several nonlinear indices of heart rate variability contained correlated informa- tion, whilst others were strongly correlated with classical linear indices. Only two nonlinear indices proved to have a prognostic value that was independent of major clinical and functional predictors, such as symptom severity, left ventricular ejection
fraction, peak V O2at cardiopulmonary exercise testing, and systolic arterial pressure.
Therefore, the quantification of nonlinear properties of heart rate variability provides important information for the risk stratification of CHF patients.
ECG parameters based on ambulatory Holter monitoring have been documented to be independent risk predictors of total mortality and progression of heart failure. Modern Holter monitoring serves as a valuable tool for investigating factors that may contribute to the mechanism of sudden death. It provides complementary informa- tion on myocardial vulnerability and the autonomic nervous system. Nevertheless, data regarding its prognostic value in the prediction of sudden cardiac death remains controversial and the positive predictive value of most Holter-based risk stratifiers is low [53].
It is unlikely that one specific ECG risk predictor could be found to predict the risk level or sudden death in a heterogeneous population of patients with chronic heart failure. Therefore, it seems that the combination of various risk markers that cover different information should be considered a better approach. Thus, in view of the breathing abnormalities presented by CHF patients, we focused on extracting information from the respiratory system.
We investigated the hypothesis that an in-depth study of the respiratory pat- tern could improve the identification of the risk level of a particular pathological disturbance and the compensatory response of the organism under pathophysiologi- cal conditions. Our previous studies [38] and [39] were focused on characterizing the frequency band that was determined by the peak of the power spectral density (PSD) associated with the envelope of the respiratory flow signal. In [79], we expanded considerably on the initial results obtained with this approach. We characterized the respiratory flow signal in CHF patients and healthy subjects using the envelope. On
the basis of autoregressive (AR) power spectral analysis of the envelope, the rele- vant discrimination band (DB) was determined from the location of the modulation frequency peak, and characterized by a number of spectral parameters.
It has been reported that the same patient might often present a mixture of breathing patterns, ranging from nonperiodic breathing (without cyclic modulation of ventilation) through to mild PB and CSR patterns [66]. Conventional spectral analysis assumes stationarity in the signal and is therefore unable to identify pattern changes. An approach which better accounts for such changes is the time-varying autoregressive (TVAR) model [80].
A study of the time-varying envelope was carried out to characterize and study dynamic changes in the respiratory flow signal in CHF patients and healthy sub- jects. The characterization involved both spectral and temporal parameters, which were extracted from the power spectrum of the respiratory flow envelope. The sta- tistical distributions of these parameters accounted for the temporal evolution of the breathing pattern [41].
To develop new quantitative parameters, our initial studies were focused on the periodicity of the respiratory pattern through the modulation of the respiratory flow signal. We characterized the relevant frequency band, which was determined by the frequency peak of the power spectral density (PSD) and related to the envelope of the respiratory flow signal [38] and [39]. Both respiratory modulation frequency and respiratory frequency are essential to the study of periodic and nonperiodic breathing patterns (PB and nPB, respectively).
The correlation function is probably the most widely used function in signal pro- cessing for quantifying the similarity of two random variables. The success of this measure depends on the assumption of Gaussian random variables, since it only con-
siders second-order statistics. Santamaria et al. recently introduced a generalization of the correlation function for stochastic processes, which was called correntropy [81] [82]. In [83] and [84], the respiratory flow signal in CHF patients with a PB and nPB pat- tern is studied through correntropy to define parameters that can improve prognosis and serve as indicators of a patient’s condition. Correntropy involves information on higher-order statistics, which can be expected to facilitate the detection of respira- tory nonlinearities that conventional second-order statistical techniques are unable to identify.