Inspiratory Muscle Training in Patients with Heart Failure

Background: Prior systematic reviews and meta-analysis addressed that inspiratory muscle training (IMT) improved inspiratory muscle weakness, cardiorespiratory fitness and quality of life similar to conventional exercise training as a first alternative in deconditioned patients with heart failure (HF) lead to a better adaptation to posterior exercise training. The heterogeneity and variability in a wide range of new studies about this topic led to the necessity of an updated and comprehensive narrative review. The present review aimed to analyze and update the most relevant studies about IMT in patients who suffer from HF. Methods: A narrative review was carried out about IMT in HF patients including 26 experimental studies divided into 21 clinical trials and 5 quasi-experimental studies identified through database searching in PubMed, Cochrane and PEDro. Results: There is enough evidence to state that IMT produces improvements in functional capacity of patients with HF. Nevertheless, there is not enough evidence to support that IMT could improve cardiovascular parameters, blood biomarkers or quality of life in these patients. Conclusions: Thus, IMT may be recommended to improve functional capacity in patients who suffer from HF; nevertheless, more evidence is needed regarding cardiovascular parameters, biomarkers and quality of life. Furthermore, mortality or HF hospitalization was not evaluated and most studies were not longer than 3 months. According to IMT protocols and study designs heterogeneity and mid-term follow-up, further investigations through high-quality long-term randomized clinical trials should be performed to achieve systematic reviews and meta-analysis to support strong evidence for IMT in HF patients.


Introduction
Worldwide, heart failure (HF) syndrome affects up to 23 million persons [1]. Furthermore, HF epidemic produces a key impact on quality of life, functional capacity and aging, as well as a high economic burden in the health system. HF may be considered as a multifactorial systemic disease involving structural, neuro-humoral, cellular and molecular mechanisms, which may be activated as a network in order to maintain physiological functioning. These complex and coordinated processes lead to an overload of the ventricles, an increased sympathetic-adrenal activity and a redistribution of the circulation, resulting in a complex clinical syndrome [2]. This syndrome may consequently produce an alteration of ventricular filling with or without reduction of the ejection fraction related to dyspnea, fatigue, exercise intolerance and peripheral and/or lung edema [1,3]. Indeed, HF syndrome is commonly divided into heart failure with reduced ejection fraction (HFrEF) and heart failure with preserved ejection fraction (HFpEF). Patients with an ejection fraction equal or greater than 50% were diagnosed with HFpEF, meanwhile patients with ejection fraction between 41% and 49% were alternatively diagnosed as HFrEF or HFpEF [1]. Patients with HF suffer from high morbidity and mortality rates, common hospitalizations and poor quality of life [4]. According to the New York Heart Association (NYHA) functional classification, HF severity may be classified as class-I without limitations or symptoms, class-II with slight limitations or symptoms during physical activity, class-III with important limitations to physical activity and class-IV with HF symptoms at rest, being exercise limitation considered as a main focus intervention in HF patients [5].
According to this consideration, exercise limitation may play a key role in HF due to an increased degree of exercise intolerance was associated with an unfavorable prognosis [6]. Patients who suffer from HF often experience an increased respiratory pattern and dyspnea during physical activity [6,7]. Despite the typical HF pathophysiological sequelae, there is not a clear relationship between cardiac function (i.e., left ventricular ejection fraction, left ventricular volumes and cardiac output) and exercise tolerance [6]. Indeed, hemodynamic abnormalities were initially considered as the main reason for these symptoms due to the ineffectiveness of the heart to increase cardiac output and pulmonary and systemic venous pressures. Increasing evidence supports a muscle hypothesis which suggested a deterioration of skeletal muscle as a source of HF symptoms [8]. Inspiratory muscle weakness in patients who suffer from HF seems to occur in a greater extent than lower limb musculature weakness [7,9,10]. This skeletal muscle atrophy may occur secondary to reduced cardiac output and tissue hypoxia, inflammation, increased systemic catabolism and prolonged immobilization, which may induce metabolic, structural, autonomic and functional changes in skeletal muscle [11]. These changes lead to protein degradation, increased levels of inflammatory cytokines (myokines), a change from slow-twitch (type-I) to fast-twitch (type-II) muscle fiber, a reduction in the number of mitochondria, impaired oxidative metabolism and early acidosis. Consequently, a reduction in muscle resistance, activation of afferent reflexes (meta-reflex) and a sustained increase in sympathetic-adrenal activity are presented. In addition, ventilation alterations increase fatigue and dyspnea as well as decrease aerobic capacity [11,12]. Thus, HF has been commonly associated with inspiratory muscle weakness, and the reduction of this inspiratory muscle weakness could have the potential to improve many secondary HF effects [7,13]. Inspiratory muscle weakness was associated with an increase in muscle meta-reflexes, which may play a key role in the clinical status of patients who suffer from HF [14].
Regarding muscle meta-reflex, this reflex may be considered as a blood pressure regulator, cardiac output, and regional distribution of muscle blood flow, involving chemically sensitive receptors located in the muscle parenchyma, which are activated by metabolites during muscle contraction [15,16]. Muscle afferent fibers of the meta-reflex are mainly comprised by unmyelinated group-IV neurons, whose receptors are chemically sensitive to metabolites produced by skeletal muscle contraction [15][16][17]. The specific type of metabolites that may activate the meta-reflex remains controversial. Some specific metabolites such as lactic acid, potassium, adenosine, arachidonic acid, diprotonated phosphate, prostaglandins or hydrogen ions have been proposed in order to activate this meta-reflex [17,18]. The efferent response secondary to meta-reflex activation may be an increased sympathetic nerve activity that could constrict the systemic vessels and increase blood flow in the contracted or active muscle, while evoking cardiac ionotropic and chronotropic effects to increase cardiac output. Therefore, this meta-reflex provokes a sympathetic-adrenal response that raises blood pressure during exercise and allows redistribution of muscle blood flow and volume, including respiratory muscles [15].
Thus, this meta-reflex may occur in respiratory muscles as a result of fatigue during an effort that may lead to metabolic sub-products accumulation, triggering the activation of receptors and consequently the meta-reflex of these muscles [15,16,[18][19][20][21]. This meta-reflex activates the sympathetic nervous system response, generating a peripheral vasoconstriction, a decrease in blood flow of skeletal muscle perfusion, an increase of exercise-induced fatigue and a redistribution of blood flow to the respiratory muscles in order to maintain their function [15,19,[21][22][23]. This increase in skeletal muscle fatigue may lead to decreased exercise tolerance and muscle strength development [19,21]. Furthermore, respiratory muscles meta-reflex can lead to an increase of heart rate and blood pressure as well as a reduction of blood flow in the renal and mesenteric arterial vessels. Simultaneously, these respiratory muscles can be influenced by both meta-reflex and chemo-reflex [18].
In addition to the meta-reflex, chemo-reflex may be considered as one of the main mechanisms to control the ventilatory and autonomic response secondary to changes in arterial oxygen, carbon dioxide and pH concentrations. Indeed, central chemoreceptors, which are located on the ventral surface of the medulla (medulla oblongata), seem to respond primarily to variations in CO 2 partial pressure (PCO 2 ) in arterial blood flow. Peripheral chemoreceptors, through type-1 glomus cells which are located in the common carotid artery and in the aorta artery, with afferences to the respiratory center which are also located in the bulb and in the solitary tract nucleus, mainly seem to respond under variations in the O 2 partial pressure (PO 2 ) in arterial blood flow [18,22]. An increase in the sympathetic nervous system activity may be secondary caused to central or peripheral chemo-reflex. Consequently, heart rate and blood pressure increases are produced, contributing to the development of hyperventilation during exercise carried out by patients with HF, showing an increase of the respiratory muscles' activity [6,22]. Meta-reflex may be considered as a powerful activator of the central chemo-reflex. According to Ribeiro et al. [22], inspiratory musculature weakness increases the peripheral chemo-reflex. Thus, this weakness of respiratory muscles may excessively increase the sensitivity to chemo-reflex by both central and peripheral responses in patients with HF, causing peripheral chemo-reflex activation at a lower threshold. This condition may lead to a sustained increase in sympathetic nervous system activity, allowing secondary adrenergic vasoconstriction and increases in right and left ventricular afterloads, being sympathetic hyper-activation considered as a key predictor of HF mortality. Therefore, inspiratory muscle strength reduction may increase the chemo-reflex and meta-reflex sensitivity in patients who suffer from HF, which could be related to reduced functional capacity and exercise intolerance [22,24]. Thus, inspiratory muscle strength may modulate meta-reflex [15,16,[18][19][20][21] and chemo-reflex [18,22], which could lead to modify respiratory system alterations and systemic symptoms like fatigue, functional capacity or quality of life [22,24]. Indeed, respiratory system alteration may be considered as one of the main factors that limit exercise capacity in patients who suffer from HF, secondary to impaired perfusion and/or ventilation [6,[25][26][27][28]. Factors which limit perfusion under HF condition may include poor right ventricular performance, elevated pulmonary arterial pressure and high pulmonary vascular resistance [25,28,29]. Indeed, inspiratory musculature weakness may be considered as one of the key factors that can limit ventilation in patients who suffer from HF [26,27]. Inspiratory muscle weakness is presented in 50% of HF patients [30] and contributes to a poor prognosis in these patients [22]. This condition is diagnosed if the maximum inspiratory pressure (PImax) is lower than 70% compared to the normalized values according to patients' age and sex [7,30] or PImax is lower or equal to 60 cmH2O [31]. PImax may be related to the dyspnea perception during daily activities and serve as a prognostic indicator in patients who suffer from HF [23,24,32,33]. Fatigue and dyspnea symptoms usually suffered by patients with HF may be partially secondary to respiratory muscles weakness [7,21,24,27,[34][35][36], due to respiratory muscle strength reduction, which could require a greater PImax fraction during respiration. Therefore, patients may experience greater dyspnea intensity secondary to increase of PImax fraction used during respiration [37]. This weakness of the respiratory musculature is also associated with a decrease in tidal volume, which may increase the ventilation-dead space ratio, increasing the ventilation-perfusion mismatch during exercise in patients who suffer from HF. In addition, an increase in the correlation between ventilation and CO 2 is generated as a key prognostic indicator in patients who suffer from HF [21,27,36,37]. Finally, the inspiratory muscles weakness promotes their fatigue, leading to an early activation of the metabolon-receptors in these muscles with possible histological changes of respiratory muscle fibers [22].
Thus, respiratory muscle weakness is accompanied by histological changes. Indeed, biopsies of the respiratory muscles performed in patients with HF have shown a lower percentage of type-IIx and type-IIa muscle fibers and a higher percentage of type-I muscle fibers, compared to healthy individuals [11,[38][39][40], being these modifications different from those observed in the limbs skeletal muscles [11]. Although the proportion of type-I fibers is usually increased in the respiratory muscles, an atrophy of these fibers has been also found under HF [21,30,40]. Furthermore, a higher percentage of type-I fibers is related to a greater oxidative enzymatic activity in these patients. These changes may be probably induced by myogenic regulatory factors linked to the increased sustained effort of ventilation. This adaptation facilitates an increase in respiratory endurance, with a parallel decrease in the maximum strength and muscular power of these muscles, leading to other compensatory mechanisms to maintain respiratory function [24,38].
With regard to other mechanisms that influence respiratory function in addition to the inspiratory muscle weakness of patients with HF, their respiratory activity is increased due to respiratory muscles having to work against increased resistive and elastic loads in these patients [6]. Elastic load is increased as a consequence of increased pulmonary tissue stiffness due to competition between the lung and heart tissue for the intrathoracic space (i.e. cardiomegaly), congested pulmonary and/or bronchial vascular flow as well as pulmonary interstitial edema. This increased resistive load may lead to lung congestion with a limitation of expiratory flow and sustained hyperventilation [6,7,21], which may result in increased oxygen and blood flow requirements of the active respiratory muscles. Given that HF may be often accompanied by a limitation in the response of cardiac output to exercise, the appearance of fatigue may be produced in an early way [6]. Furthermore, HF patients present a maladaptive respiratory pattern with a shorter expiratory time and a longer inspiratory time due to inspiratory muscles weakness and reduced inspiratory resistance [41]. This condition may also cause sympathetic-adrenal hyperactivity, being this condition a predictor of mortality under HF [22], because parasympathetic activity in the sinus node is decreased during inspiration and parasympathetic activity is upregulated during expiration, which hinders the functional capacity of these patients [42].
In this regard, Yamada et al. [21] demonstrated in patients who suffered from HF the relationship between respiratory muscles weakness and functional capacity limitation during a 6-minute walking test, regardless of suffering from a restrictive respiratory pattern or lower limb skeletal muscle weakness. Furthermore, inspiratory muscle weakness was related to different functional classes of NYHA, showing the highest inspiratory force for patients with class-I condition and the lowest inspiratory force for patients with class-IV, inspiratory muscle training (IMT) being considered as a possible key rehabilitation intervention to improve this symptomatology in patients suffering from HF [24,41].
Cardiac rehabilitation is a well-studied and comprehensive rehabilitation program that has been proven to improve functional capacity in heart failure patients. Nevertheless, these cardiac rehabilitation programs are currently underutilized [43]. IMT may serve as a useful alternative that may be more amenable to HF patients' participation. Furthermore, IMT may serve populations excluded from cardiac rehabilitation, such as patients unable to perform exercise, being an interesting treatment option for clinicians [33,44].
According to these antecedents, prior systematic reviews and meta-analysis have addressed that IMT improved inspiratory muscle weakness, cardiorespiratory fitness and quality of life similarly to conventional exercise training as a first alternative in deconditioned patients with HF, leading to a better adaptation to posterior exercise training [33,44]. In 2013, Smart et al. [44] analyzed 11 controlled trials of IMT in chronic HF patients including data on 287 patients divided into 148 patients who received IMT and 139 patients who were assigned to sham or sedentary control groups. In 2014, Montemezzo et al. [33] analyzed 9 randomized controlled clinical trials including 240 patients and comparing IMT with controls or sham interventions. To date, there is a lack of literature reviews about IMT in patients with HF and several experimental studies, including clinical trials, quasi-experimental studies and clinical cases, have been published. The heterogeneity and variability on a wide range of new studies about this topic led to the necessity of an updated and comprehensive narrative review.
The present review aimed to analyze and update the most relevant studies about IMT in patients who suffer from HF.

Study Design
A narrative review was carried out following the applicable recommendations of the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) criteria [45]. This narrative review was performed in order to update the available data from prior systematic reviews about IMT in patients who suffer from HF [33,44] but including different studies types, such as clinical trials, quasi-experimental studies and case-series, published up to January 2020.

Search Strategy
Database searching process was carried out during January 2020. PubMed, Cochrane and PEDro were the used databases for this process. Restrictions used in this database search were experimental studies carried out only in humans with access to full texts using the following search strategy ("Inspiratory muscle training" OR "respiratory muscle training") AND "heart failure" in the title of abstract from studies which were written in Spanish or English languages.

Selection Criteria
Inclusion criteria comprised experimental studies, such as clinical trials, quasi-experimental studies or clinical cases, including HF who received IMT intervention. Exclusion criteria comprised non-experimental studies as well as systematic reviews or meta-analyses and studies which included patients with other conditions in addition to HF such as strokes, lung hypertension, "Fontan circulation" or mitral valve alterations, among other pathologies. All possible outcome measurements were accepted for review because our aim was to update all information and assess variable experimental studies in HF patients who were treated with IMT in an isolated form or in combination with other interventions [33,44].

Data Extraction
Studies characteristics such as sample size, socio-demographic data and baseline measurements as well as training protocol, including duration, frequency, intensity, IMT device and protocol were registered. These data were divided into 3 tables including characteristics of the randomized clinical trials without other interventions (Table 1a), characteristics of the randomized clinical trials with other interventions (Table 1b) and characteristics of the quasi-experimental studies and clinical cases (Table 1c).
Data extraction comprised study citation, group and sample sizes, outcome measurements with pre-and post-intervention means ± standard deviations (SD), statistical significance (p-values), and additional information about outcome measurements procedure was provided when it was necessary. These data extraction was divided into several tables including inspiratory muscle strength and resistance (Table 2a), lung function (Table 2b), dyspnea (Table 3a), fatigue (Table 3b), functional class and capacity (Table 4), strength in limbs (Table 4b), parameter related to VO 2 (Table 5), VE/VCO 2 and VE (Table 6), cardiovascular parameters (Table 7), biomarkers ( Table 8) and quality of life (Table 9), according to the updated literature and prior reviews [33,44].

Flow Diagram
From 218 records identified through the searching process, 192 records were removed due to duplicates and exclusions (Supplemental Table S1), and finally, 26 studies were included in narrative synthesis ( Figure 1).  inspiratory muscular resistance in patients who suffer from HF [46].

249
IMT may be considered as a training method by workloads applied during inspiration [33]. IMT 250 can be performed in three different ways by an inspiratory load threshold device, a resistive load 251 threshold device and an isocapnic hyperpnea. First, IMT using inspiratory load threshold devices is 252 applied using an inspiratory pressure to cause the valve openness and thus allow air flow to pass 253 during inspiration. Second, IMT using resistive loading devices is applied by several holes of 254 different diameter that provide resistance to inspiration using a smaller diameter of the hole to 255 provide a greater resistance. Third, IMT using an isocapnic hyperpnea device can be only performed 256 in a well-equipped respiratory physiology laboratory and is applied maintaining a certain level of 257 ventilation in the form of volitional hyperpnea for 12 minutes mean while CO2 is added to the 258 inspired air in order to maintain isocapnia in arterial blood flow [47].

259
To date, 26 articles have examined the effects of IMT in patients with HF (

Inspiratory Muscle Training
Inspiratory muscle weakness observed in patients with HF is reversible. Thus, inspiratory muscle training (IMT) is one of the key interventions for the improvement of the strength and inspiratory muscular resistance in patients who suffer from HF [46].
IMT may be considered as a training method by workloads applied during inspiration [33]. IMT can be performed in three different ways by an inspiratory load threshold device, a resistive load threshold device and an isocapnic hyperpnea. First, IMT using inspiratory load threshold devices is applied using an inspiratory pressure to cause the valve openness and thus allow air flow to pass during inspiration. Second, IMT using resistive loading devices is applied by several holes of different diameter that provide resistance to inspiration using a smaller diameter of the hole to provide a greater resistance. Third, IMT using an isocapnic hyperpnea device can be only performed in a well-equipped respiratory physiology laboratory and is applied maintaining a certain level of ventilation in the form of volitional hyperpnea for 12 min mean while CO 2 is added to the inspired air in order to maintain isocapnia in arterial blood flow [47].
Other experimental studies included 4 quasi-experimental studies [48,58,64,65] and one clinical case [59]. Two of the quasi-experimental studies compared IMT with control or healthy groups. In addition, citations, sample characteristics and training protocol explanation for these experimental studies were detailed in Table 1c.
In most of all these studies, the intervention group only performed IMT, and the control group used a simulated IMT (no load), a low intensity fixed load or only received education [20,30,31,35,48,49,51,[54][55][56]58,63,[65][66][67][68][69]. The remaining 9 studies also included a physical training program in addition to performing IMT in the intervention group [50,52,53,57,[59][60][61][62]64]. IMT programs were mainly differentiated into four variables, such as the percentage of the PImax or the sustained maximum inspiratory pressure (SMIP), the duration of the training session, the weekly frequency of IMT and the total duration of the training program. PImax percentages varied from 20% to 60% in the intervention groups, with 30% being the most used PImax percentage. The SMIP ranged from 15% to 60%. Total training periods ranged from 4 to 12 weeks; durations for each IMT session varied from 15 to 30 min, and frequencies varied from two daily sessions to three weekly sessions.
Lung function was measured in nine studies [30,35,51,53,59,60,64,67,68] and demonstrated that IMT did not seem to be a useful intervention to improve significantly pulmonary function because statistically significant differences (p < 0.05) for some parameters were only shown in 3 studies [35,51,68]. Forced vital capacity (FVC), forced expiratory volume in the first second (FEV1), FEV1/FVC ratio, maximum expiratory flow and vital capacity were the parameters used to determine lung function. FVC increased statistically significant in 2 studies [35,68]. FEV1 showed statistically significant differences in only 1 study carried out for Laoutaris et al. [68]. FEV1/FVC ratio and maximum expiratory flow were statistically and significantly increased in the study performed by Bosnak-Guclu et al. [51]. Vital capacity did not show statistically significant differences (p > 0.05) regarding studies that assessed this parameter [59,64]. Training intensity equal or greater than 40% showed statistically significant improvements for one or more lung function parameters in 3 studies [35,51,68] from the total of 5 studies [35,51,53,67,68] that used this training intensity (Table 2b).     Indeed, other important data to highlight are provided in the following 3 studies. Mancini et al. [64] observed an improvement (p < 0.05) in maximum voluntary ventilation without variation in the ti/ttot ratio. Chiappa et al. [48] showed an increase in the thickness of the diaphragm in addition to a hypertrophy of the diaphragm that was associated with a statistically significant increase (p < 0.001) in PImax. Finally, Laoutaris et al. [67] reported an increase in the inspiratory volume (p < 0.001).
In summary, IMT may be considered as a useful therapy to improve the strength and endurance of the respiratory muscles. Authors encouraged researchers to assess lung function after performing IMT in future studies dividing the groups into one that performed IMT at an intensity between 20% and 30% of the PImax or SMIP and the other group at an intensity of ≥40% of the PImax or SMIP in order to observe if higher intensity improvements in lung function may be achieved.

Dyspnea
IMT effects on dyspnea were measured in 15 studies [30,31,35,[51][52][53][54]61,[63][64][65][66][67][68][69] and demonstrated that IMT may be a useful treatment to reduce dyspnea at both rest and during exercise performance. All these studies showed a significant improvement (p < 0.05) in dyspnea except for the study carried out by Johnson et al. [66]. Most studies used the Borg scale to assess dyspnea [30,31,52,53,[64][65][66][67][68][69], and only 2 of them did not observe a statistically significant difference (p > 0.05) in the dyspnea sensation [64,66]. Nevertheless, Mancini et al. [64] did not observe statistically significant improvements for dyspnea; patients walked a greater distance with the same dyspnea sensation and concluded that HF patients were able to increase working charge without dyspnea sensation increase. Three studies used the Modified Medical Research Council (MMRC) scale [51,54,61] showing statistically significant improvements with respect to the control group, except for the study carried out by Marco et al. [54], which did not show statistical significance. Two additional studies measured the dyspnea index [35] and the Mahler transition index [63]. Weiner et al. [35] revealed statistically significant differences. Nevertheless, Martínez et al. [63] did not reported these improvements. Laoutaris et al. [68] correlated as statistically significant (p < 0.01) the decrease in dyspnea with an increase in the distance covered in the 6 min walking test (Table 3a).
Fatigue was only analyzed in 2 studies, and both studies showed a fatigue decrease after IMT intervention [51,61], although only the study carried out by Hossein Pour et al. [61] showed statistically significant differences (p < 0.05) with respect to the control group. Both studies used the Fatigue Severity Scale (FSS) to assess fatigue [51,61] (Table 3b).
Thus, IMT may reduce the sensation of dyspnea in these patients, both at rest and during exercise performance. Authors encourage researchers to carry out further studies to obtain more evidence regarding the decrease in fatigue after the use of IMT.
11.8 ± 0.6 (6) 11.5 ± 0.6 (6)   [66] did not report statistically significant differences (p > 0.05). Two of the studies presented both control and IMT groups performing a general physical training program and did not show statistically significant differences between groups [50,57]. The improvement in the 6-minute walk test was related to the improvement in PImax (p < 0.01) and SMIP (p < 0.05) [63] (Table 4a).
Eight studies evaluated the time to exhaustion at a given exercise intensity [30,52,53,64,[66][67][68][69]. In all these studies, an improvement was shown in favor of the IMT group (Table 4a), except for 3 studies [53,66,69]. Quadriceps strength was measured in 5 studies [51,52,59,60,62], showing statistically significant improvements (p < 0.05) in favor of the IMT group for all reported studies. In these research reports, three studies [52,60,62] showed that IMT group also performed strength training, and the control group did not carry out this training or trained at a lower intensity. Strength for upper limbs was examined in three studies [54,59,60], and significant improvements in favor of the IMT group were only shown in the study carried out by Kawauchi et al. [60], which used a control group with strength training at a lower intensity than in the IMT group. Furthermore, Marco et al. [54] did not find a statistically significant correlation (p > 0.05) between the improvement of PImax or PEmax and the grip strength (Table 4b).
The study carried out by Bosnak et al. [51] was the only study that evaluated and observed a significant improvement (p < 0.05) in balance in the IMT group. This fact may be probably due to the increase in the diaphragm thickness after performing a relatively short IMT period (4 weeks) [48]. Thus, greater strength and resistance of the inspiratory muscles may contribute to better balance, since the diaphragm and other breathing muscles may be responsible for maintaining both breathing and balance [27,48,51,52,55]. Therefore, peripheral and central adaptations of IMT may improve the mechanisms responsible for respiration, ventilation and balance and subsequently produce a synergy between these processes [27].
Four studies [52,53,60,61] assessed functional capacity by the NYHA scale. Form these 4 studies, the study carried out by Hossein Pour et al. [61] was the only research that observed between-groups statistically significant differences (p < 0.05), although the other 3 studies [52,53,60] showed an IMT intra-group improvement but did not show statistically significant differences (p > 0.05) with respect to the control group. These results may be due to isolated IMT improved NYHA functional class, although IMT in conjunction with physical exercise did not support additional benefits (Table 4a).
Weiner, 1999 IMT n = 10 458 ± 29 2 562 ± 32 2 <0.01 p-value *: showed intra-group statistical significance. p-value **: showed between-groups statistical significance. p-value ***: p-value of intervention group vs. control at the same time point. * If it is specified, we refer to meters; meanwhile, feet were specified by the abbreviation ft. 1 : Expressed as minutes ( ) and seconds ("). 2    In conclusion, IMT may improve the distance covered in walking tests and endurance to maintain a certain exercise. Nevertheless, higher quality studies are needed using isolated IMT in one group to demonstrate significant improvements in limb strength or functional NYHA class. Regarding endurance to maintain a certain exercise level until exhaustion, current evidence is ambiguous to assess the effect of IMT on limb-strength increase. Higher quality studies would be necessary considering the following two aspects. Firstly, examining whether isolated IMT could improve this parameter since only two studies [51,54] used isolated IMT. Secondly, observing if IMT could provide additional benefits to certain training methods assessing if both IMT and control groups could obtain differences using a different IMT protocol and the same training protocol.

Metaboreflex Activity
Three studies [20,48,55] analyzed the IMT effects on the meta-reflex of respiratory muscles, demonstrating a statistically significant improvement (p < 0.05) in blood flow in the limb muscles at rest or during exercise due to a decrease in limb vascular resistance. IMT significantly increased the ventilatory load required to provoke the peripheral vasoconstriction mediated by respiratory muscles meta-reflex. This fact may be due to a greater resistance to fatigue in these muscles, reducing the accumulation of metabolites that triggered the meta-reflex activation [19,20,48,55,70]. Thus, IMT may be associated with an effect of inspiratory muscle release by increasing the inspiratory muscles thickness, strength and diaphragmatic muscular aerobic capacity. This effect was associated with less accumulation of metabolites and reduced peripheral vasoconstriction, resulting in increased peripheral blood flow and greater exercise tolerance [19,27,48]. According to this decrease in meta-reflex, Mello et al. [55] observed that IMT reduced muscle sympathetic activity improving sympathetic and vagal modulation of the cardiovascular system in patients with HF. Furthermore, this decrease in sympathetic-adrenal activity could be also due to a reduction in chemo-reflex since IMT could improve oxygen saturation. In addition, Moreno et al. [20] showed that respiratory muscles meta-reflex decreased after IMT. Thus, an oxygen saturation improvement of the intercostal and forearm muscles was observed.
Therefore, IMT may reduce inspiratory muscles' meta-reflex in patients who suffer from HF.
The circulatory power was calculated as the product of the VO 2 peak by the maximum systolic blood pressure [30,50,52,53,69]. Two of these studies [53,69] did not show statistically significant differences (p > 0.05). Oxygen absorption efficiency slope (OUES) values were analyzed in two studies [49,50] showing a statistically significant improvement (p < 0.05) of these values in the IMT group. In addition, Laoutaris et al. [69] observed a significant improvement in oxygen pulse (VO 2 /HR) in the IMT group, but there were not statistically significant differences with respect to the control group.
In conclusion, IMT may increase peak VO 2 and circulatory power. Regarding OUES, CP and oxygen pulse, IMT could improve these parameters, although a larger number of studies should be carried out. Finally, IMT did not seem to be a useful therapy to improve VO 2 anaerobic threshold.
Therefore, there is a lack of evidence in order to affirm that IMT may reduce the VE/VCO 2 slope.
Four studies assessed LVEF after performing IMT and did not show significant differences (p > 0.05) between groups [52,53,56,67], although intra-group statistical significance (p < 0.05) was shown in two studies [52,53]. The left ventricular diameter at the end of systole or diastole was also evaluated [52,53,56,67]. Only two studies carried out by Laoutaris et al. [52] and Adamopoulos et al. [53] showed an intragroup significant decrease in left ventricular diameter at the end of diastole observed in both IMT and control groups, being the only two studies that combined IMT with strength training. Thus, these improvements may be due to the training program. Finally, the left atrial volume index was measured in two studies [56,57], showing a significant improvement in the IMT group after 6 months according to Palau at al [57].
Heart rate variability was assessed in two studies [55,69], determining a significant increase (p < 0.05) in this parameter in the IMT group according to Mello et al. [55]. Laoutaris et al. [69] justified the absence of improvement because the evaluation was carried out for 24 h, and the patients in that period carried out physical exercise. This fact supposed an increase in the activity of the sympathetic system and therefore a decrease in the variability of heart rate.
Resting diastolic blood pressure was measured in two studies and neither of them showed any intra-group or between-group statistical significance (p > 0.05). Peak systolic blood pressure was assessed only in one study [47], which did not show any statistically significant difference. Furthermore, resting systolic blood pressure was assessed in three studies [52,55,69], which did not show any statistically significant difference.
In summary, there is a lack of evidence to support that IMT may improve any cardiovascular parameter analyzed in the present review. Further and high-quality studies about cardiovascular parameters are required in the near future.

Biomarkers
The existence of an abnormal immune response seems to play an important role in the HF pathogenesis and progress, including overexpression of pro-inflammatory cytokines such as alpha tumor necrosis factor (TNF-α), interleukin-6 (IL-6), and soluble apoptosis mediators such as soluble Fas (sFas) and soluble Fas ligand (sFasL) [68]. Other biomarkers such as the N-terminal pro-brain natriuretic peptide (NT-proBNP) or serum carbohydrate 125 antigen (CA125) may be related to the HF severity, and C-reactive protein (CRP) may be used as an indicator of systemic inflammation [53,69,72].
Six studies [53,54,56,57,68,69] assessed the IMT effects on blood biomarkers, being the NT-proBNP, CA125 and CRP considered as the most studied biomarkers (Table 8). Five studies [53,54,56,57,69] evaluated the effect of IMT on NT-proBNP, although only the study carried out by Adamopoulos et al. [53] showed a significant difference (p < 0.05) in favor of IMT, highlighting that this study was the only study that included an aerobic training program. The effect of IMT on CA125 was analyzed in two studies [56,57], showing the absence of significant differences between groups (p > 0.05). CRP was evaluated after IMT in three studies [53,54,68], determining a significant difference in favor of IMT according to Adamopoulos et al. [53]. In addition, Laoutaris et al. [68] analyzed IMT modifications on TNF-a, IL-6, sFas and sFasL, and sTNF-RI, although only significant improvements were observed in favor of IMT for sTNF-RI and sFas values. According to Marco et al. [54], improvements were shown for renal function in favor of the sham IMT group, since the glomerular filtration rate was increased (p = 0.007), and creatine was decreased (p = 0.003).
In conclusion, IMT did not seem to produce key changes in blood biomarkers, although additional studies about sTNF-RI and sFas could be interesting for clinicians and researchers.
In summary, IMT improved quality of life under MLwHFQ evaluations but not for SF-36 assessments. Thus, it would be advisable to carry out a study to determine the existence of differences between MLwHFQ and SF-36 in HF patients.

Limitations and Future Studies
Despite data extraction and statistical significances being included in all tables for intra-group and between-group comparisons, some PRISMA checklist recommendations [45], such as methodology quality assessment, risk of bias, heterogeneity analyses and forest plots, were not followed due to insufficient information in part due to the heterogeneity of analyses. The heterogeneity of our study sample hindered our capacity to carry out a systematic review or meta-analysis. Thus, an extensive narrative review was performed including all available scientific evidence about this topic. Indeed, this narrative review analyzed all available experimental studies, including clinical trials, quasi-experimental studies and clinical cases, about IMT in patients with HF. According to all PRISMA recommendations [45], future studies should be carried out following systematic reviews and meta-analysis for specific outcome measurements in order to update strong recommendations to apply IMT interventions in patients who suffered from HF [33,44].

Conclusions
To date, there is enough evidence to state that IMT produces improvements in respiratory muscle strength and endurance, dyspnea sensation at rest and during exercise, distance covered in gait tests, resistance during exercise, meta-reflex of respiratory muscles, peak VO 2 and circulatory power. Nevertheless, there is no evidence to determine that IMT could improve VO 2 anaerobic threshold , cardiovascular parameters and blood biomarkers. In addition, there is a lack of evidence to confirm that IMT may improve quality of life due to controversial findings between MLwHFQ and SF-36 measurements, lung function, VE/VCO 2 , OUES, oxygen pulse, NYHA functional class and limb strength. Finally, mortality or HF hospitalizations were not evaluated, and most studies were not longer than 3 months. According to IMT protocols and study design heterogeneity and mid-term follow-up, further investigations through high-quality long-term randomized clinical trials should be performed to achieve systematic reviews and meta-analysis to support strong evidence for IMT in HF patients.