Electrocardiographic and other Noninvasive Hemodynamic Markers in Decompensated CHF Patients

Acutely decompensated chronic heart failure (adCHF) is among the most important causes of in-hospital mortality. R-wave peak time (RpT) or delayed intrinsicoid deflection was proposed as a risk marker of sudden cardiac death and heart failure decompensation. Authors want to verify if QR interval or RpT, obtained from 12-lead standard ECG and during 5-min ECG recordings (II lead), could be useful to identify adCHF. At hospital admission, patients underwent 5-min ECG recordings, obtaining mean and standard deviation (SD) of the following ECG intervals: QR, QRS, QT, JT, and T peak–T end (Te). The RpT from a standard ECG was calculated. Patients were grouped by the age-stratified Januzzi NT-proBNP cut-off. A total of 140 patients with suspected adCHF were enrolled: 87 (mean age 83 ± 10, M/F 38/49) with and 53 (mean age: 83 ± 9, M/F: 23/30) without adCHF. V5-, V6- (p < 0.05) RpT, and QRSD, QRSSD, QTSD, JTSD, and TeSD p < 0.001 were significantly higher in the adCHF group. Multivariable logistic regression analysis demonstrated that the mean of QT (p < 0.05) and Te (p < 0.05) were the most reliable markers of in-hospital mortality. V6 RpT was directly related to NT-proBNP (r: 0.26, p < 0.001) and inversely related to a left ventricular ejection fraction (r: 0.38, p < 0.001). The intrinsicoid deflection time (obtained from V5-6 and QRSD) could be used as a possible marker of adCHF.


Introduction
Chronic heart failure (CHF) is still a serious, unresolved problem that undermines the economic sustainability of all national healthcare systems [1,2]. The massive hospital admission rate for acute, decompensated CHF (dCHF) poses a major burden on healthcare systems, being costly and negatively impacting patients' quality of life. Obviously, the current COVID-19 pandemic has dramatically challenged and stressed all health systems, worsening the already critical situation of chronic patients, increasing hospital admissions as well as mortality and morbidity [3][4][5][6]. Thus, the remote monitoring of CHF patients seems to represent the most easy and cost-effective solution, utilized in order to adapt and customize drug therapy to prevent new decompensations and hospitalizations [6]. Then, the identification of new non-invasive, low-cost, simple, and transmissible markers to monitor these patients seems mandatory. Recently, some ECG markers able to identify adCHF patients have been studied; in particular, markers of short-term repolarization variability appeared promising to individuate patients with high levels of NT-proBNP [7,8], 2 of 12 which is the gold standard for the diagnosis of acute CHF patients [9] and high mortality risk [7,8,[10][11][12]. In addition, it was reported that an increase in the R-wave peak time (R p T) or, more specifically, in the intrinsicoid deflection time (R p T obtained only in the precordial unipolar leads) [13] was a risk factor for sudden cardiac death [14,15] or heart failure events [16]. The aim of this study was to prove that a simple, inexpensive, non-invasive marker as the intrinsicoid deflection time was suitable for diagnosing adCHF and to verify that short-term RpT variability, expressed as five minutes of mean standard deviation, was consequently affected in these patients.

Materials and Methods
A total of 140 consecutive patients were enrolled at the admission to our Geriatric Department from January 2019 to July 2022 due to dyspnea or other symptoms of decompensated chronic heart failure (CHF), diagnosed according to the 2021 European Society of Cardiology heart failure guideline [9]. Patients have been grouped in two different groups (subjects with or without adCHF) according to the age-stratified value of the NT-proBNP cut-off proposed by Januzzi [17]. In particular, the cut-offs were 450, 900, and 1800 pg/mL, respectively, for patients less than 50, between 50 and 75, and older than 75 years old [17]. Exclusion criteria were: complete left or right ventricular branch block, intra-ventricular conduction delay (120 ms), or paced QRS; and active pneumonia, SARS-CoV2 infection, or high degree of cognitive impairment which leaves the patients unable to sign the written informed consent for the participation in the study. All patients had stable previous clinical conditions at home with NYHA class II-III, and all of them were in the IV NYHA functional class at the time of enrollment. At the study run-in, each patient underwent a full clinical history, physical examination, standard ECG, transthoracic echocardiogram, and simultaneously 5 min of noninvasive hemodynamic evaluation using bioimpedance cardiography (Physioflow, Manatec Biomedical, Folschviller, France) and a single-lead (II lead) ECG recording. Furthermore, at this time, a blood sample for NT-proBNP dosage (Alere Triage Analyzer, Alere, San Diego, CA, USA) as well other serum variables was collected. ECG signals were acquired and digitalized with a custom-designed card (National Instruments, Austin, TX, USA) at a sampling frequency of 500 Hz. Measurements used for the ECG segment and interval analysis were detected automatically using a classic adaptive first derivative/threshold algorithm and template method [18]. Our research group designed and produced software for data acquisition, storage, and analysis with the LabView program (National Instruments, Austin, TX, USA). An expert cardiologist (GP) checked the different ECG intervals and segments, automatically marked using the software, and, when needed, manually corrected the mistakes [18][19][20][21][22]. The mean and standard deviations ( SD ) of the following study intervals and segments on five minutes ECG recordings were recorded: mean of R-R intervals (RR); the mean of Q-R intervals (QR): from q to the peak of R waves (R-wave peak time); the mean of Q-R-S intervals (QRS): from q to S waves; the mean of Q-T intervals (QT): from q to end of T waves; and the mean of ST end segments (ST) from S to the end of T waves. RR SD was reported only in sinus rhythm subjects and without premature supraventricular or ventricular beats. In patients with a sinus rhythm, a power spectral analysis with an autoregressive algorithm was also used and evaluated [8,[23][24][25][26][27]. Next, the total area under the spectra was defined as the total power (TP) and it was equivalent to the variance in the RR; very low-frequency power (VLF) was the area between 0 and 0.04-hertz equivalents (Hz); low-frequency power, those between 0.04 and 0.15 (Hz); and high-frequency power (HF), those between 0.15 and 0.40 hertz. We also calculated the ratio between LF and HF (LF/HF). In addition, we transformed the absolute power in the LF and HF normalized units (nu) [23][24][25][26][27][28]. The R p T was calculated manually using an electronic caliper on a 12-lead ECG (Esaote, model: MyCardioPad/XL 12 channels, Florence, Italy) available for the analysis (paper speed was 25 mm/s and calibration 10 mm/mV). Inter-observer measurement error was avoided by using measurements made by the same trained operator. The intra-observer and measurement errors of R p T were defined.

Statistical Analysis
Data were reported as mean ± standard deviations or as median and interquartile ranges for normal and skewed distribution data. The categorical variables were reported as frequencies and percentages. We compared non-normally (as evaluated using the Kolmogorov-Smirnov test) and normally distributed data, respectively, using the Mann-Whitney test and unpaired t-test. Categorical variables were compared using the χ 2 test. Uni-and multi-variable forward (A. Wald) stepwise logistic regression analyses were used to determine the association between mortality and the studied variables (QT and Te intervals). Stepwise multiple regression analysis was used to determine possible relationships between the studied variables. Receiver operating characteristic (ROC) curves were used to determine the sensitivity and specificity of the studied parameters predictive of decompensated CHF (positive NT-proBNP with Januzzi cut-off) and areas under the curves (AUCs), and 95% confidence intervals (CI) were calculated to compare the diagnostic accuracy. For statistical analysis, we used SPSS-PC+ (SPSS-PC+ Inc., Chicago, IL, USA). We considered statistically significant p values < 0.05.

Results
217 patients were eligible, 23 patients were not included because the quality of standard ECG or short-term ECG recordings was not optimal for the study, and 54 patients were not included for intra-ventricular conduction delay or complete left or right branch block. Therefore, we conducted the study on 140 patients (Table 1) regarding R p T and short-term ECG recording data; at the same time, we obtained hemodynamic data only from 58 subjects due to the availability of the necessary tool (from January 2021).
According to the NT-proBNP Januzzi cut-off, 87 patients were grouped as adCHF and 53 as patients without decompensated CHF. During the hospitalization, a total of thirty-two patients died (overall mortality rate, 23%), nineteen (14%) died for bronchopneumonia and respiratory failure (one patient for COVID-19 pneumonia), nine for terminal heart failure (6%), two for fatal myocardial infarction (1%), two for sudden cardiac death (one for sustained ventricular tachycardia and ventricular fibrillation; one for acute cor pulmonale secondary to massive embolism). The overall cardiovascular mortality rate was 9%. Regarding the deceased patients, 29 out of 32 (91%) were in decompensated CHF group (p < 0.001) ( Table 1).
β-blockers and calcium antagonists did not have any influence on the R p T; in fact, even comparing the groups for treatments, no significant difference between patients with or without antiarrhythmic drugs was found.

Discussion
The major findings of this study were that the intrinsicoid deflection time, obtained from the unipolar precordial leads (V 5-6 ) of a standard ECG, was higher in patients with adCHF. Secondly, these patients showed an increased QR SD ; in fact, we observed a higher level of R p T in short-term temporal dispersion, obtained in lead II, in patients with adCHF. To avoid misunderstandings, it is important to reiterate the definition of R p T and intrinsicoid deflection time: the first one regards the interval between q and R waves in all ECG leads, the second one represents the same ECG interval but obtained in the precordial leads. Moreover, the delayed intrinsicoid deflection lasts equal to or more than 50 ms in the V 5 and V 6 leads [13,36]. Normally, this interval lasts less than 50 ms in healthy subjects, but it tends to grow with the increase in cardiac hypertrophy [36], and it was proposed as a risk marker of sudden cardiac [14,15] or heart failure acute events [16]. Our data showed that this interval is affected by the ejection fraction more than by the left ventricular hypertrophy degree; so, it could be considered as a left ventricular function marker rather than a heart hypertrophy marker. This finding confirmed that this electrical marker could be used to monitor CHF patients.
From an electrophysiological point of view, the R p T coincides with the upstroke velocity (phase 0) of the ventricular myocyte action potential, and it is characterized by the rapid inflow of sodium ions in the ventricular myocardiocytes. Normally, the duration of this phase is of a few milliseconds, but in dilative or hypertrophic myocardiopathy, it is associated with electrical myocardial remodeling, an increase in action potential duration and its temporal inhomogeneity, as well as with an increase in R p T and intrinsicoid deflection velocity. The electrophysiological basis, studied in different animal models, could involve the sodium, calcium, and potassium channels, but also structural changes such as the reduction in gap junctions, fibrosis, muscle fiber disarray, and cellular size [36]. Therefore, multifactorial mechanisms could be involved in RpT prolongation, but the neurohumoral activation in CHF decompensated patients could probably play the lead role. High catecholamine levels, inducing a downregulation of β-adrenergic myocardial receptors, could reduce chrono/dromo/inotropy, causing an increase in the R p T too. In the present study, we observed a decrease in all spectral components and total short-term variability, expressed as both TP and RR SD ; this behavior was observed in advanced CHF, in the animal experimental model [19,26], and the clinical study, too [24,27,[37][38][39][40][41].
Another interesting point was that, despite the drug therapy with potential effects (β-blockers, calcium channel blockers, and antiarrhythmics) on the R p T in these patients, no possible influence of therapy on the studied patients was detected, probably due to the receptors' downregulation, frequently seen in these patients; therefore, these findings confirmed the handling of intrinsicoid deflection and the short period of R p T temporal dispersion as markers of heart failure decompensation. Finally, an increase in the short period temporal dispersion of different ECG intervals or segments was predictive of CHF decompensation, as previously observed [7,8]. On the contrary, the possible use of delayed intrinsicoid deflection or R p T as a sudden cardiac death marker in these patients was not confirmed, as our study reported only two cases of sudden cardiac death.
Finally, the bioimpedance analysis was useful to individuate patients with adCHF, specifically thanks to two parameters: LVEF BIO and EDFR. The first was indirectly calculated according to the Capan equation [42], and the second was an index of diastolic dysfunction. In fact, this last index was positively related to the peak Doppler echocardiographic early diastolic velocity [43]. In conclusion, these non-invasive hemodynamic indexes could find a prominent place in the possible remote monitoring of these patients, but new prospective studies are needed.

Conclusions
The continuous aging of the population and the recent pandemic have increasingly highlighted the need to promote health and quality of life, especially for patients with chronic diseases. Heart failure, especially in its more advanced stages, is subject to frequent exacerbations that put patients' lives at risk and weigh heavily on health systems. The phenomenon of the "revolving door" in fact determines the increasingly frequent need for hospitalization due to the rapid succession of acute decompensation events. The possibility of remotely monitoring chronic patients should be applied to all subjects with chronic heart failure, in order to guide therapeutic choices and avoid exacerbations. Non-invasive technologies specifically suited for elderly and frail patients with CHF are not yet widely available. In this study, we attempted to evaluate adCHF using simple, inexpensive, easily interpretable, and repeatable electrical markers. The hope is to improve the monitoring and therapy of outpatients with CHF and reduce the high rate of rehospitalization. Present and future research should point towards this new frontier and beyond the challenges of the coming years.

Limitations
A current limitation of the present study is represented by the paucity of patients, despite the sample size having been statistically calculated and allowing for a reliable statistical significance of the results. It will be desirable to plan a multicenter study with a larger case series to confirm the data obtained.
Furthermore, the enrolled population is basically composed by elderly patients. This aspect is therefore not attributable to an arbitrary choice made by the authors, but to the characteristics of the population under study. In fact, heart failure in the advanced stage with frequent episodes of decompensation is prevalent in the elderly population; so, the mean age of both the heart failure group and the control group remains above 80 years. In our opinion, this aspect does not limit the results of the study; on the contrary, it highlights a positive trend in the data obtained which, while requiring further studies, lay the foundations for the use of these clinical and electrocardiographic parameters in telemedicine and telemonitoring, specifically useful in this category of fragile patients. A separation into groups by age was followed due to the existence of different cut offs of NT-proBNP depending on the age of the patients according to the Januzzi criteria [17]. The authors do not exclude that with other classifications, it would be possible to obtain other results. At the moment, however, it seems logical to follow this type of subdivision.
Another point is that the intrinsicoid deflection may be subject to numerous "disturbing elements" and biases; nevertheless, we did not observe substantial differences among the patients in the study and the control group (i.e., drugs intake). These data therefore reinforce the observation and the diagnostic and prognostic power of the variables studied.
In conclusion, the data obtained in this study, although preliminary, are in line with the existing literature, confirming the possibility of creating a multiparametric score that is easy and immediate to use and rather cheap, all in order to stage the risk of acute decompensation or the mortality of patients affected by decompensation.