Early Cardiac Rehabilitation Improved Prognosis in Patients with Heart Failure Following Acute Myocardial Infarction

Background: Cardiac rehabilitation (CR) has been shown to improve exercise intolerance and QoL, and minimize re-hospitalizations in patients with congestive heart failure (CHF). However, studies on early CR in patients with acute myocardial infarction (AMI) who developed CHF following percutaneous coronary intervention (PCI) are rare. The purpose of this study is to evaluate the effectiveness of early CR on patients with CHF after AMI following PCI. Methods: Two hundred thirty-seven patients who developed heart failure after AMI following PCI were enrolled. Patients were divided into heart failure with reduced ejection fraction (HFrEF) group and heart failure with mid-range ejection fraction (HFmrEF) group. Of which, 78 patients who accepted a two-week CR were further divided into two subgroups based on major adverse cardiovascular events (MACE). Key cardio-pulmonary exercise testing (CPX) variables that may affect the prognosis were identied through the comparison of the cardio-respiratory tness (CRF). Results: Early CR signicantly reduced cardiac death in patients with HFrEF, and reduced re-hospitalization in patients with HFmrEF after AMI (P <0.01). Serum potassium and CR ratio were independent risk factors for MACE in patients with both HFrEF and HFmrEF after AMI. In the CR group who developed MACE, there were more diabetics (P=0.035), with higher serum potassium (P=0.043), and lower P ET CO 2 at VT (P=0.016). P ET CO 2 at VT was an independent risk factor for re-hospitalization. The incidence of re-hospitalization was signicantly lower when the P ET CO 2 at VT was greater than 33.5mmHg (P=0.03). Conclusions: Early CR reduced the incidence of MACE in patients with heart failure after AMI following PCI. The P ET CO 2 at VT is an independent risk factor for re-hospitalization, and could be used as a key evaluating hallmark for early CR in patients who developed heart failure after AMI. per minute at anaerobic threshold,E-VCO 2 : Exercise Carbon dioxide production, △ VCO 2 : Margin of Minute ventilation Carbon dioxide production, VE/VCO 2 slope: Minute ventilation/Carbon dioxide production,R-P ET CO 2 : Rest Partial pressure of end-tidal carbon dioxide, P ET CO 2 at VT: Partial pressure of end-tidal carbon dioxide at anaerobic threshold, △ P ET CO 2 : Margin of Partial pressure of end-tidal carbon dioxide, MACE: major cardiac events, IQR: range.


Background
Congestive heart failure (CHF) is a major cause of mortality and morbidity and the end pathophysiological condition of many cardiovascular diseases [1] . One of the leading causes of CHF is myocardial infarction.
Percutaneous coronary intervention (PCI) signi cantly decreased the mortality in patients with acute myocardial infarction (AMI) [2] . However, CHF continues to develop in some patients before or soon after PCI.
Exercise intolerance, represented as decreased capacity to perform physical activities with symptoms of severe fatigue and/or dyspnea, is a characteristic of CHF and associated with increased mortality and reduced quality of life (QoL) [3] . The pathophysiological mechanisms of exercise intolerance in CHF are multifactorial, involving impaired cardiac and pulmonary reserve as well as decreased respiratory and peripheral skeletal muscle function [4] . In addition to conventional treatment, many researchers have shown that cardiac rehabilitation (CR) can improve exercise intolerance and QoL, and minimize re-hospitalizations in patients with CHF [5,6] . However, studies on early CR in patients who developed CHF soon after AMI following PCI are scarce. A pilot study done by Houchen L et al. indicated that early CR could signi cantly reduce depression, enhance exercise tolerance and decrease CHF-associated hospital admission [7] . However, the study population was small and no control group was presented for comparison. In view of this, this study evaluated patients with CHF after AMI following PCI with and without CR, and compared biochemical parameters and cardio-respiratory tness (CRF), as well as long-term prognosis at 4 years follow-up.

Patient population:
From January 2015 to January 2016, patients admitted for acute ST elevation myocardial infarction with CHF following PCI were identi ed in the Department of Cardiology at the First Hospital of Jilin University. The study protocol was approved by the Institutional Review Board of the hospital.
Patients' baseline characteristics and biochemical parameters were collected from medical records. According to the 2016 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure [8] , patients were divided into heart failure with reduced ejection fraction (HFrEF) group and heart failure with mid-range ejection fraction (HFmrEF) group. In each group, patients were further divided into CR and non-CR subgroups. CR consisted 2 weeks exercise including 3 supervised regular exercise sessions per week on a bicycle [9] and 4 supervised electrical stimulation sessions per week for no regular exercise day [10] . Regular exercise session lasted for 20 minutes including warm-up and cool-down, and included three 3-minute intervals aiming at Borg 11-13 by subjective sensation separated by 2-minute recovery periods of 0 watt intensity [9] . Electrical stimulation was performed 30 min/day, 4 days per week, using a dual-channel battery-powered stimulator Elpha-II 3000 (DANMETER® A/S, Odense, Denmark). The stimulator delivered a biphasic current of 25 Hz frequency. The electrical current characteristics were set up as follows: "on-off" mode stimulus (3 s stimulation, 6 s rest), pulse width 300us, rise and fall time 1 s. The intensity of the stimulation was adjusted to produce a visible muscle contraction but not too strong to make the patients uncomfortable [10] . Adhesive electrodes were placed on both legs over the upper and lower aspects of gastrocnemius muscles, and over the upper-lateral and lower-medial portions of the quadriceps muscles. After the 2 weeks CR, patients were advised to continue individualized exercise at home for 3-4 sessions per week. Individualized exercise prescription was created based on each patient's CRF from cardio-pulmonary exercise testing (CPX) before discharge [11] . Patients who accepted the 2weeks CR were subsequently reassigned into two subgroups based on the major adverse cardiac events (MACE), namely the MACE group and the non-MACE group. The parameters of CRF between the two subgroups were compared, and the main CPX variables that may predict the prognosis of patients with CHF were identi ed.
Quanti cation of cardio-respiratory tness CPX, which is a widely accepted evaluation tool in both the United States and Europe [12,13] , was used for the assessment of CRF. The measurement of ventilatory gas exchange was used to predict prognosis of death and rehospitalization [13][14][15] . In CR patients, the oxygen consumption (VO 2 ), carbon dioxide production (VCO 2 ), minute ventilation (VE), partial pressure of end-tidal carbon dioxide (PETCO 2 ), respiratory exchange ratio (RER) and other key CPX variables were measured with submaximal graded exercise test using Cardio-respiratory instrumentation Medisoft (Made in Belgium, Model:E100000011000001 N:130619-05-1470, ) after 2-week CR. The exercise load was determined via a cycle ergometer work rate. The progressive load was 10 watts per minute during the graded exercise test. The exercise test was terminated if the patient developed any of the following subjective or objective conditions: abnormal hemodynamic or electrocardiographic exercise response, or other causes such as dyspnea,angina or lower extremity muscle fatigue.
Clinical follow-up Follow-up data was acquired through hospital records and telephone interviews which were conducted every 3 months from discharge until cardiac death or July 2019, whichever came rst. MACE including cardiac death and re-hospitalization were documented. Patients with cardiac death who lost telephone interviews were identi ed from the population registry bureau. The average duration of follow-up was 4 years.

Statistical analysis
Continuous variables were expressed as means ± standard deviation, and non-normally distributed variables were presented as medians (interquartile range). Categorical variables were expressed as numbers and percentages. Variable parameters between the groups were compared with means of one-way analysis of variance, or Mann-Whitney U test for continuous variables and chi-square test for dichotomous variables, as appropriate. In all analyses, a two-tailed p < 0.05 was considered as statistical signi cance.
Cox multivariate regression analysis, in which test indices and variables showing p < 0.05 in the univariate analysis were included, was used to determinate independent risk factors for MACE. A receiver operating characteristic (ROC) curve was used to predict the prognosis for MACE. All statistical analysis data were performed using the SPSS 19 software (IBM Corp., Armonk, NY, USA).

Results
A total of 274 patients of AMI with CHF following PCI were identi ed, after excluding 21 patients who were lost to follow-up and 16 patients who were not able to participate cardiopulmonary exercise testing (CPX) ( Table 1), 237 patients were included in the nal analyses. Of which, 55 patients had HFrEF (n = 22 in CR and n = 33 in non-CR group) and 182 had HFmrEF (n = 56 in CR and n = 126 in non-CR group). In HFrEF group, there were no signi cant differences in baseline characteristics between CR and non-CR groups except that the non-CR group patients were older with fewer smokers. In HFmrEF group, there were more male patients in non-CR group otherwise there were no signi cant differences in baseline characteristics between CR and non-CR groups ( Table 2).

Incidence of major cardiovascular events
In the HFrEF group, non-CR patients had higher MACE rate (P = 0.002) due to higher incidence of cardiac death (P = 0.002) as compared to CR patients ( Table 2, Fig. 1). In the HFmrEF group, non-CR patients had higher MACE rate (P < 0.001) due to higher incidence of heart failure (HF) re-hospitalization (P = 0.008) as compared to CR patients ( Table 2, Fig. 1).

The main CPX variables for prognosis prediction
The 78 patients who accepted the 2-weeks CR were subsequently reassigned into two subgroups based on the MACEs, namely the MACE group (n = 6) and the non-MACE group (n = 72) ( Table 7).    Compared with non-MACE group, more patients in MACE group were diabetic (P = 0.035), had higher serum potassium (P = 0.043), higher incidence of heart failure re-hospitalization (P < 0.001) and myocardial infarction (P = 0.005), and lower P ET CO 2 at VT (P = 0.016). P ET CO 2 at VT was found to have predictive value for rehospitalization. The area under the curve was 0.789 and the cut-off point was 33.5 mmHg (Fig. 3). P ET CO 2 at VT was an independent risk factor for re-hospitalization (OR = 0.635, 95% CI: 0.463-0.871, P = 0.005), as shown in Table 8. The incidence of re-hospitalization was signi cantly lower when the P ET CO 2 at VT was higher than 33.5 mmHg (P = 0.03) (Fig. 4).

Discussion
The present study is the rst retrospective study evaluating early CR in patients with heart failure after AMI following PCI. The main ndings of this study suggest that early CR was able to reduce cardiogenic death in patients with HFrEF, and reduce re-hospitalization in patients with HFmrEF after AMI. Furthermore, the intervention was safe; P ET CO 2 at VT was an independent risk factor for re-hospitalization.
In patients with HF, research suggested that exercise-based CR could improve QoL, decrease all-cause hospital admissions and HF-dependent hospital admissions in the short term (up to 12 months) and potentially reduce mortality in the long term when compared to no exercise patients [16,17] . Our study expands the previous research by showing that early rehabilitation program involving supervised regular exercise and electrical stimulation can reduce the incidence of cardiac death in patients with HFrEF, and heart failure re-hospitalization in patients with HFmrEF. Elevated serum potassium level and CR ratio were independent risk factors for cardiac death in HFrEF patients after AMI. Moreover, our study suggests that sex, age,history of stroke, and elevated serum potassium were independent risk factors for re-hospitalization in patients with HFmrEF after AMI. Thomsen et.al also reported that hyperkalemia was strongly associated with the degree of renal dysfunction and severe clinical outcomes as well as death in HF patients [18] . Houchen et.al found that a 6-week early rehabilitation program including exercise and self-management education, improved short-term exercise tolerance and depression [7] . However, the sample size was small in this study and there was no control group for comparison. With a larger study population and a non-CR control group, out results suggest that early CR was not only effective but also safe in patients with CHF soon after AMI.
Exercise intolerance is a major feature of CHF, and is associated with impaired QoL, reduced functional capacity and poor prognosis. In addition to reduced cardiac function, other causes such as reduced pulmonary reserve, impaired skeletal muscle function, etc can diversely and signi cantly contribute to the syndrome in CHF patients, and even turn into the dominant mechanisms of exercise intolerance [19] . Exercise can provide numerous bene ts for CHF patients including decreased long-term morbidity and mortality [20] , improved cardiac remodeling [21] , increased neurovascular functional competency [22] , reduced re-hospitalization and improved of cardiorespiratory capacity and QoL [1,23] . The bene ts of electrical stimulation included improving blood supply and muscle strength, as well as exercise tolerance in severe CHF patients [24,25] , so it could be offered as an alternative to bicycle training as part of a home-based rehabilitation program [10] . In our study, the re-hospitalization in patients who accepted the 2-week CR after PCI was only related to CRF, but not to serum potassium level or history of diabetes. The reason may be related to the protective effects of exercise on renal function and the improvement of glycolipid metabolism. Our previous research suggested that upregulation of nitric oxide synthases in the kidney and left ventricle may contribute, in part, to the renal and cardiac protective effects of exercise training in cardio renal syndrome in chronic heart failure rats [26] . Furthermore, exercise can reduce early diabetic nephropathy by upregulating nitric oxide synthases as well as ameliorating NADPH oxidase and α-oxoaldehydes in the kidneys of ZDF rats [27] .
CRF is now being considered as an essential variable and should be assessed in health screenings [28] . The clinical values of CRF assessment include diagnosis, functional evaluation and prognosis prediction. CPX is the most precise tool to determine exercise tolerance and considered as the reference clinical procedure for assessing CRF by quantifying peak VO 2 which represents an individuals' capacity to generate energy for strenuous exercise [28] . CHF is a systemic syndrome with the reduction of functional reserve being the outstanding characteristic. The cardiovascular impairment has a direct negative effect on other systems and organs, including the respiratory, renal and neuromuscular systems. CPX is de ned as "gold standard" for the CRF of patients with cardiovascular disease, the clinical diagnosis assessment and prognosis prediction can be achieved from direct measurement of VO 2 , VCO 2 and VE [29] .  [32] . Ferreira et al. noted that the cutoff point of VE/CO 2 slope ≧ 43 would be an appropriate hallmark for heart transplantation determination [33] . Chua et al. found that CHF patients with VE/VCO 2 slope > 34 were at very high risk for death and re-hospitalization [34] . The 2012 EACPR/AHA scienti c statement endorsed that peak VO 2 and VE/VCO 2 slope are the most studied CRF variables in CHF patients and both indicated signi cantly independent prognostic value [11] . For patients under medical treatment, a peak VO 2 < 10.0 ml/kg/min and a VE/VCO 2 slope ≧ 45 exist at the same time would indicate a very poor prognosis over the following 4-year [11] . It should be noted that, in order to achieve the prediction accuracy of peak VO 2 value on CHF, maximal exercise (at least RER > 1.05) should be achieved during the test [29] . Others evaluating the long term prognosis by VE/VCO 2 slope in CHF, and reported that it was an excellent independent value, even better than peak VO 2 , and could be achieved only from sub-maximal exercise [35,36] . Similarly our results indicated that CR patients with VE/VCO 2 slope < 36 had a good cardiovascular prognosis.
However, it is di cult to achieve a maximal test in most CHF patients due to the exercise intolerance. The 2016 EACPR/AHA updated the scienti c statement, and felt that it is important to note that VO 2 at VT holds broad applicability in the context of assessing the capacity [37] . The VO 2 at VT has also been indicated as a hallmark for the prognosis prediction prior to surgery [38,39] . Furthermore, we also showed that VO 2 at VT is a signi cant prognostic marker for AMI patients in whom a VO 2 at VT < 10.5 ml/kg/min indicated a poor long term prognosis [40] . The P ET CO 2 both at rest and in exercise have been found to be positively correlated with the prognosis of systolic heart failure [41] . Abnormalities in the VE/VCO 2 slope and P ET CO 2 in patients with HCM have been thought Page 17/26 to enhance pulmonary pressures [31,41] . In the present study, we noted that P ET CO 2 at VT was also a predictor of re-hospitalization for patients with CHF after AMI, previous study showed that a lower P ET CO 2 was associated with a poor prognosis [40] .
Our previous study demonstrated that VO 2 at VT was an independent risk factor for cardiovascular disease prognosis and could be used as an evaluating hallmark for Phase I cardiac rehabilitation in patients with STEMI after PCI [40] . However, the P ET CO 2 at VT is an independent risk factor for re-hospitalization in patients with heart failure after AMI. The reasons of the difference may be as follows: rst, patients with STEMI after PCI may not have reduced ejection fraction and severe pulmonary dysfunction after PCI operation, a smaller amount of P ET CO 2 is an indicator of less CO 2 production in the body and/or pulmonary arterial perfusion, or in other words, the cardiac output [42] . The difference in P ET CO 2 might be attribute to the difference in infarct location. VO 2 at VT is determined not only by the degree of the infarct area but also the peripheral oxygen utilization e ciency, and could be an independent risk factor for the prognosis in patients with STEMI after PCI. Second, ventilation is regulated by the sensitivity of respiratory chemo-receptors and the ergo re ex in skeletal muscles. The sensitivity of respiratory chemo-receptors increases when the sympathetic nerve is activated and/oracidosis occurs. These conditions often occur in HF patients [43] . These subjects experience shortness of breath throughout mild to vigorous activity. While in insu ciently expansion and with increased dead space between artery and alveolus, diffusion of CO 2 is less, hence, P ET CO 2 decreases. The re-hospitalization associated with exercise intolerance in patients with CHF are multifactorial, including impaired cardiac and pulmonary reserve, and decreased respiratory and peripheral skeletal muscle function, all of them can diversely and signi cantly contribute to the decrease in P ET CO 2 . The combined aerobic/resistance/inspiratory training in patients with CHF has been shown to produce positive changes in left ventricular structure and function, which provided additional bene ts in both peripheral and diaphragmatic muscle function, dyspnea, cardiopulmonary exercise parameters and QoL [44] .
The limitations of this study include: 1) this is a retrospective study in which patients were not randomly assigned to CR or non-CR group, thus selective bias cannot be avoided; 2) patients in the non-CR group were not assessed for CRF using CPX before discharge; 3) the number of MACE in the early CR group was relatively small. Therefore, further research is needed to con rm the in uence of CPF parameters on prognosis prediction and the accuracy of cut-off point value.

Conclusions
In conclusion, an early CR decreased the incidence of cardiovascular events in patients with CHF after AMI following PCI. Elevated serum potassium and CR ratio were independent risk factors for MACE in patients not only with HFrEF but also with HFmrEF after AMI. The P ET CO 2 at VT was an independent risk factor for rehospitalization, and could be used as a key evaluating hallmark for early CR in patients with CHF after AMI.

List Of Abbreviations
AMI: acute myocardial infarction CHF: congestive heart failure CR: cardiac rehabilitation The Kaplan-Mayer curves of MACE-free survival in the HFrEF group. CR: cardiac rehabilitation, NCR: non cardiac rehabilitation, MACE: major cardiac events.