Effects of β2-receptor stimulation by indacaterol in chronic heart failure treated with selective or non-selective β-blockers: a randomized trial

Alveolar β2-receptor blockade worsens lung diffusion in heart failure (HF). This effect could be mitigated by stimulating alveolar β2-receptors. We investigated the safety and the effects of indacaterol on lung diffusion, lung mechanics, sleep respiratory behavior, cardiac rhythm, welfare, and exercise performance in HF patients treated with a selective (bisoprolol) or a non-selective (carvedilol) β-blocker. Study procedures were performed before and after indacaterol and placebo treatments according to a cross-over, randomized, double-blind protocol in forty-four patients (27 on bisoprolol and 17 on carvedilol). No differences between indacaterol and placebo were observed in the whole population except for a significantly higher VE/VCO2 slope and lower maximal PETCO2 during exercise with indacaterol, entirely due to the difference in the bisoprolol group (VE/VCO2 31.8 ± 5.9 vs. 28.5 ± 5.6, p < 0.0001 and maximal PETCO2 36.7 ± 5.5 vs. 37.7 ± 5.8 mmHg, p < 0.02 with indacaterol and placebo, respectively). In carvedilol, indacaterol was associated with a higher peak heart rate (119 ± 34 vs. 113 ± 30 bpm, with indacaterol and placebo) and a lower prevalence of hypopnea during sleep (3.8 [0.0;6.3] vs. 5.8 [2.9;10.5] events/hour, with indacaterol and placebo). Inhaled indacaterol is well tolerated in HF patients, it does not influence lung diffusion, and, in bisoprolol, it increases ventilation response to exercise.

Pulmonary function test with lung diffusion measurements. Standard pulmonary tests were performed according to the American Thoracic Society criteria 11 . The normal predicted values for FEV 1 and VC were those reported by Quanjer et al. 12 . Lung diffusion for carbon dioxide (DL CO ) and for nitric oxide (DL NO ) were simultaneously measured in the standard sitting position through the single-breath technique, with a breath-hold time of 4 seconds (MS-PFT analyzer, Jaeger Masterscreen, Hoechberg, Germany). Membrane diffusion (D M ) subcomponent was calculated dividing DL NO by 1.97, while capillary volume (Vc) was estimated from DL CO and D M through Roughton and Foster's formula 13 . Alveolar volume (V A ) was measured by helium decay slope during single-breath constant expiratory flow measurement.
CPET. Maximal CPETs were performed on a cycle ergometer (Ergo 800S, Sensor Medics, Yorba Linda, CA) applying a ramp protocol personalized for each patient, designed to reach maximum exercise in about 10 ± 2 minutes 14 . All patients performed at least one familiarization procedure before the protocol run-in. Patients breathed into a mass flow sensor through a mouthpiece connected to a saliva trap and wearing a nasal clip or through a facial mask, according to their preference and to facial and dental conformation. Ventilation (VE), oxygen consumption (VO 2 ), and carbon dioxide production (VCO 2 ) were measured breath by breath (V-max 2900 metabolic cart, Sensor Medics, Yorba Linda, CA). HR and 12-lead EKG were monitored continuously, hemoglobin saturation was recorded by an ear oximeter, and blood pressure was monitored with a cuff sphygmomanometer every 2 minutes. Anaerobic threshold (AT), VE/VCO 2 slope, and VO 2 vs. work relationship were measured following a standard methodology 15 .
Nocturnal cardiorespiratory monitoring (NCM). Sleep respiratory behavior was recorded by a portable system (Embletta x100, SapioLife S.r.L., Monza, Italy) with simultaneous and continuous recording of respiratory flow and snoring by a nasal cannula, thoracic and abdominal respiratory effort by strain gauges, and oxygen saturation (SaO 2 ) by a finger digital oximeter. Apneas were defined as a reduction in the amplitude of respiratory flow signal below 10% of the baseline value for at least 10 seconds, while hypopneas were defined as a reduction of respiratory flow between 10 and 50% of the baseline for at least 10 seconds associated with an oxygen desaturation of at least 3%. Apneas were considered of central origin when the interruption in respiratory flow was associated with absence of thoracic and abdominal respiratory effort, obstructive if a respiratory thoracic or abdominal activity was present during the stop in respiratory flow, and mixed when an initially central apnea turned into obstructive in its terminal phase 16 . Hypopneas were not further classified. Apnea-hypopnea index (AHI) was defined as the number of apneas and hypopneas per hour of time in bed (defined as the time spent between light switch off and switch on, in recumbent position and without major body movements). Central apnea, mixed apnea, obstructive apnea, and hypopnea indexes were calculated with the same method as AHI. Study design. All consecutive patients fulfilling inclusion and exclusion criteria and who gave their written informed consent to the study were enrolled, and they performed a maximal CPET to identify a personalized ramp protocol. The day before starting the experimental drug administration, a 24-hour Holter EKG recording and an NCM were performed. On day one of the study (V1), procedures other than Holter EKG and NCM were performed as described above. Holter EKG data were immediately evaluated to exclude subjects with severe ventricular arrhythmias. After completion of all tests, patients were blindly randomized 1:1 to indacaterol or placebo treatment for a two-month period, planning the first follow-up examination after 3 days and the following every 15 days to check for drug compliance and tolerability and for adverse events. After two months of treatment (V6), patients performed all the study procedures. After a 14-day wash-out of the experimental drug, patients returned to the research laboratory (V7) to repeat the same study procedures and to restart treatment with the experimental drug according to a cross-over design: patients taking placebo in the V1-V6 period shifted to indacaterol treatment, and vice versa. After another two-month treatment period with examinations planned at the same intervals as in the V1-V6 period, all patients performed the study procedures again (V12), and the study was considered completed (Fig. 1). At V2, V3, V4, V5, V8, V9, V10, and V11, patients underwent clinical evaluation, and compliance to study protocol was assessed. In order to maintain the double blindness, an external firm produced identical capsules and identical packaging for both placebo and indacaterol. Blister packs were identified by the expression "period 1" (administered between V1 and V6) and "period 2" (administered between V7 and V12), containing placebo or indacaterol, following a predetermined randomization list obtained with a pre-specified software (PMX CTM 3.2/(c) Propack Data GmbH, NerPharMa) with 25 blocks (block size = 4). Both patients and physicians were blinded to the content of the blister packs.
The primary endpoints of the study were the changes in lung diffusion, expressed by DL CO , after indacaterol treatment compared with placebo treatment in the whole population and in the bisoprolol and carvedilol groups, and the safety of indacaterol in the whole population as evaluated by clinical, EKG, 24-hour Holter EKG, and blood chemistry parameters.

Statistical analysis.
A population sample of 60 patients (30 for each β-blocker group) was planned in order to detect a difference in the primary endpoint (DL CO difference between indacaterol and placebo) of at least 2 ± 1.5 ml/min/mmHg with a power >95%. Continuous variables are presented as mean ± SD, while non-normally distributed variables are presented as median and interquartile ranges (IQR). The analyses to evaluate differences between the indacaterol and the placebo group were performed using paired t-test for AB/ BA cross-over design 17 . Subsequently, a subgroup analysis was implemented to assess the potential difference between drug and placebo groups within and between the two β-blockers, yet again with paired t-test for AB/ BA cross-over design. All calculations were computed with the aid of the SAS software package (Version 9.4 SAS Institute Inc., Cary, NC).

Compliance with Ethical Standards
Research involving human participants and/or animals. This research involves humans. This article does not contain any studies with animals performed by any of the authors. Informed consent. Informed consent was obtained from all individual participants included in the study.

Results
All eligible patients referring to the HF Unit of Centro Cardiologico Monzino between September 2015 and September 2016 were consecutively evaluated for participating in the study, and 48 patients gave their written informed consent to the study protocol and begun the study treatment and procedures. One patient died during follow-up because of cardiac arrest as a consequence of HF worsening during the placebo treatment period. Four patients experienced serious adverse events: one during placebo (hospitalization because of systemic inflammatory disease), 2 during indacaterol treatment (paroxysmal atrial fibrillation and acute bowel obstruction, respectively) and one in the wash-out period (transient cerebral ischemia few days after the end of the indacaterol treatment period). One patient reported a non-severe adverse event during placebo (dyspnea). Only one adverse event (the atrial fibrillation episode) was considered to be possibly related to the experimental drug. Globally, 4 out of 6 patients (2 on placebo and 2 on indacaterol) who reported adverse events discontinued the experimental drug and did not complete the study procedures (one because of death and three having withdrawn their informed consent after discussion with the medical team about the risk of study prosecution), while 2 patients continued the research protocol. Patient enrollment was prematurely interrupted after the inclusion of 48/60 patients because of the unavailability of placebo doses 12 months after the beginning of the study. In the end, 44 patients completed all study procedures through the whole study period and were included in the analysis, 27 of them were treated with bisoprolol and 17 with carvedilol. The baseline characteristics of the entire study population are reported in Table 1. No significant differences were observed between patients receiving carvedilol and bisoprolol except for a significantly higher rest HR and an almost significantly lower LVEF in the carvedilol group (respectively HR 70 ± 10 vs. 63 ± 10 bpm, p < 0.03, and LVEF 31.2 ± 6.8 vs. 34.6 ± 4.6%, p = 0.051) (Table  in Supplementary Information).
Indacaterol-related effects were assessed comparing active drug vs. placebo (V6 and V12, Fig. 1). In Table 2, data related to treatment safety are reported, derived from clinical evaluation, 12-lead EKG, blood samples, and 24-hour Holter EKG in the whole population. No differences were observed in any of the investigated parameters after indacaterol treatment in comparison to placebo, with the only exception of maximum HR recorded during 24-hour Holter EKG, which appeared to be significantly higher with indacaterol.
In Tables 3-5, the effects of indacaterol treatment on lung mechanics, lung diffusion, exercise, and nocturnal respiratory behavior are reported in the entire population and separately in the carvedilol and bisoprolol groups. No significant difference was observed in the whole study population between indacaterol and placebo treatments for any of the explored parameters, except for a significant increase in VE/VCO 2 slope (Fig. 2) and a reduction in maximal P ET CO 2 after indacaterol, and a significant, but clinically negligible, longer snoring time with indacaterol treatment. Similarly, no significant effect of indacaterol vs. placebo was observed in the carvedilol group, with the only exception of a significant reduction in the hypopnea index. Conversely, in the bisoprolol group, a trend was observed towards higher vital capacity and FEV 1 and a significantly greater ventilatory response to exercise (higher VE/VCO 2 slope) and lower maximal P ET CO 2 after indacaterol as compared to placebo. A significant inverse correlation was observed as well between the changes in maximal P ET CO 2 and VE/VCO 2 slope after indacaterol compared to placebo in the whole population (R −0.48, p < 0.002).

Discussion
In spite of the limited number of patients enrolled, some conclusions about indacaterol treatment can be drawn from our study: (a) indacaterol administration was safe in chronic HF patients treated with a β-blocker; (b) indacaterol induced a minor improvement in lung mechanics but only in bisoprolol-treated patients; (c) indacaterol had no effects on alveolar-capillary membrane diffusion regardless of β-blocker treatment; (d) indacaterol increased the ventilatory response during exercise but only in the bisoprolol group with no effects on overall exercise performance; (e) no major indacaterol-induced effects were observed on sleep quality, except for a reduction in hypopneas in the carvedilol-treated group. COPD is a frequent comorbidity burdening HF prognosis, whose treatment is often challenging because of the possible detrimental side effects of most bronchodilator therapies 18 . A particular concern regards β 2 -stimulating agents, whose use has been frequently associated with tachycardia, hyperkinetic arrhythmias, and worsening HF in cardiac patients 3,4,19 . A major result of our study is that indacaterol, a long-acting topical selective β 2 -agonist widely used for COPD treatment 20 , proved to be safe in a population of HF patients with moderate-to-severe reduced LVEF in stable clinical conditions and on optimized pharmacological treatment. Indeed, the incidence of both serious and minor adverse events was overall trivial and similar during treatment with indacaterol and placebo. Moreover, no significant differences were observed between indacaterol and placebo in cardiac arrhythmias, blood pressure, resting HR, BNP, or MLWHFQ score. Only a slightly, albeit significantly, higher value of  www.nature.com/scientificreports www.nature.com/scientificreports/ maximum HR during Holter EKG recording was detected with indacaterol treatment in comparison with placebo, but with similar mean and minimum HR in the 24-hour period.
In HF patients, an impairment in lung mechanics has been reported, particularly in patients with severe HF 21 . The patients we studied had moderate HF with a limited reduction in standard spirometry parameters. Nevertheless, a minor improvement was observed with indacaterol, but only in bisoprolol-treated patients. We recognize, however, that the clinical relevance of this finding is questionable, although a greater effect might be hypothesized in patients with more severe HF, who have a greater impairment of lung mechanics.
The main hypothesis of our study was that indacaterol could increase lung diffusion thanks to a more efficient removal of alveolar fluids, at least in HF patients whose alveolar β 2 -receptors are still active (i.e. not blocked by carvedilol). Indeed, the role of β 2 -receptors in modulating alveolar fluid clearance has been widely documented both in healthy subjects and in HF patients 5,7,8,[22][23][24] . Our data do neither confirm nor deny this hypothesis. As a matter of fact, indacaterol did not change lung diffusion in comparison with placebo, either in patients treated with bisoprolol or in patients treated with carvedilol. Indeed, DL CO , DL NO , D M , and Vc were not significantly different in indacaterol vs. placebo, independently of the β-blocker used. This observation is actually in line with those obtained by other groups, whose results were published after the beginning of our study, on the effects of acute inhalation of β 2 -agonists. In healthy subjects, Taylor NE et al. did not observe any significant change in D M after acute administration of albuterol 21 . In a further study on HF patients, the same group observed a reduction in lung water content after acute albuterol administration, as evaluated by computed tomography imaging combined with DL CO /DL NO -derived capillary volume, but without any difference in DL CO , DL NO , D M , or Vc between before and after drug administration 10 Table 4. Comparison of cardiopulmonary exercise test parameters after treatment with placebo or indacaterol in the whole study population and in the cardvedilol and bisoprolol groups. VO 2 = Oxygen Consumption; VCO 2 = Carbon Monoxide production; VE = Ventilation; Vt = tidal Volume; RR: Respiratory Rate; HR = Heart rate; RER = Respiratory Exchange Ratio; AT = Anaerobic Thresold, PETCO 2 = maximal end-tidal CO 2 partial pressure.
administration in COPD patients on DL CO , even though a protective effect against lung diffusion injury after acute fluid overload challenge was detected 9 . These negative observations, however, do not necessarily imply that the stimulation of alveolar β 2 -receptors does not influence alveolar-capillary membrane diffusion conductance in HF patients. In fact, even if a mild degree of bronchodilation was observed in our population in bisoprolol-treated patients after indacaterol, there is no evidence that the drug actually reaches the alveolar compartment after topic administration. Indacaterol, as well as other β 2 -agonists, has been conceived to reach bronchioles, where it exerts its bronchodilation activity, and only marginally the alveolar tissue, where it can be absorbed in the systemic circulation and exert undesired side effects. Considering the huge area of the alveolar compartment, it is possible that the amount of drug able to stimulate the alveolar β 2 -receptors was too low to induce any detectable effect on lung diffusion. Ideally, a β 2 -agonist able to reach to a relevant extent the alveolar compartment, which to our knowledge does not exist nowadays, should be tested to further investigate the possible benefit of alveolar β 2 -receptor stimulation on lung diffusion in HF patients. Finally, indacaterol proved to increase the ventilatory response to exercise, as demonstrated by a significantly higher VE/VCO 2 slope 25 , in patients treated with bisoprolol but not in those treated with carvedilol. This observation was at first unexpected. Indeed, ventilation depends on VCO 2 production, dead space/tidal volume (VD/Vt) course − which in turn depends on alveolar ventilation/perfusion matching − and arterial pCO 2 set point, which depends on chemoreceptor sensitivity, according to the following equation: VE = k × VCO 2 /[PaCO 2 × (1 − VD/ Vt)] 26 . Peak VCO 2 was not significantly modified by indacaterol either in bisoprolol-or in carvedilol-treated patients. VD/Vt could not be reliably calculated during exercise or at the peak of exercise in our study because of the lack of arterial samples during the tests. From a theoretical point of view, indacaterol could induce an increase in exercise VD, mostly in patients treated with a cardioselective β-blocker such as bisoprolol, assuming that bronchodilation induces a recruitment of non-perfused alveoli. However, if this were the case, an increase in maximal ventilation at the peak of exercise during indacaterol treatment, to a higher extent in the bisoprolol www.nature.com/scientificreports www.nature.com/scientificreports/ group, should have been observed, but it was not. Moreover, on one hand, the worsening of ventilation/perfusion mismatch during exercise is usually observed only in very severe HF patients 25 , whereas our population consisted mostly of patients with moderate left ventricular systolic dysfunction and in stable clinical conditions, and, on the other hand, the increase in FEV 1 observed after indacaterol in the bisoprolol group (suggesting a mild degree of bronchodilation) was quantitatively negligible. An alternative, more convincing interpretation is that indacaterol increased ventilatory response to exercise by increasing chemoreceptor sensitivity to CO 2 through a stimulation of β 2 -receptors located on chemoreceptors themselves. The observed reduction of P ET CO 2 in parallel with VE/ VCO 2 increase is in line with this hypothesis. Consistently with this interpretation, this effect was only observed in bisoprolol-treated patients, and not in patients whose β 2 -receptors had been blocked by carvedilol. Indeed, chemoreceptor activity is widely modulated by hormonal and nervous stimuli, and β 2 -receptor stimulation or blockade has proved to influence chemoreceptor sensitivity, particularly in HF patients 8,27 . Indeed, there are several differences between bisoprolol and carvedilol, and carvedilol induces a more complete sympathetic block. First of all, bisoprolol is highly cardioselective, having therefore no or only trivial effects, at the usual dosages, on β 2 -receptors located in the lung, while carvedilol equally blocks β 1 and β 2 -receptors. Moreover, carvedilol blocks both β and α-receptors, it does not upregulate cardiac β-receptors, it has central sympatholytic activity, and it is classified as a β-arrestin-biased agonist 1,28,29 . These actions could explain the different effect on chemoreceptor sensitivity, and consequently the greater ventilatory response to exercise obtained with indacaterol in the bisoprolol group. The absence of any relevant effect of indacaterol on sleep apnea, regardless of the β-blocker treatment, is not surprising, since our patients had very limited sleep abnormalities and the presence of sleep abnormalities was not among the study inclusion/exclusion criteria. It is possible, but totally unproven, that results could have been different in patients with more severe HF and sleep abnormalities. Study limitations. Some limitations of the study need to be acknowledged.
(1) The main limitation of this study is that it was prematurely interrupted because of placebo shortage one year after the beginning of the recruitment. Consequently, the number of cases in the two β-blocker groups were different. Accordingly, more data are needed to define some of the study endpoints. However, the primary study endpoint, i.e. the effects of indacaterol on alveolar-capillary gas diffusion, is clearly negative, and it would have remained so even if all cases had been studied. (2) Carvedilol and bisoprolol groups were not created by a randomization system. This is a source of potential bias, as the clinical choice of the β-blocker is usually driven by comorbidities and in particular by the presence or absence of lung disease. However, severe lung disease was an exclusion criterion, and no significant difference was observed in patients' clinical characteristics between carvedilol and bisoprolol groups, but for a slightly higher HR and lower LVEF in bisoprolol (see table in Supplementary Information). (3) It is totally unknown whether and how much of the inhaled drug gets to the alveoli and whether other forms of inhalation (ultrafine, different inhalator, etc.) could produce different responses.

Conclusions
Indacaterol, a long-acting, highly β 2 -selective adrenoceptor agonist widely used for the treatment of COPD, proved to be safe in moderate-to-severe HF patients in stable clinical conditions treated with a β-blocker. An increase in ventilatory response to exercise after indacaterol was observed only in patients treated with a β 1 -selective β-blocker (bisoprolol), suggesting a role of β 2 -receptor stimulation in the regulation of the ventilatory drive. Finally, our study failed to demonstrate any effect of indacaterol treatment on lung diffusion. The inability of the indacaterol molecule to reach the alveoli and to stimulate β 2 -receptors located on the alveolar side of epithelial cells is a possible explanation.