Autonomic and respiratory consequences of altered chemoreflex function: clinical and therapeutic implications in cardiovascular diseases

Abstract

The importance of chemoreflex function for cardiovascular health is increasingly recognized in clinical practice. The physiological function of the chemoreflex is to constantly adjust ventilation and circulatory control to match respiratory gases to metabolism. This is achieved in a highly integrated fashion with the baroreflex and the ergoreflex. The functionality of chemoreceptors is altered in cardiovascular diseases, causing unstable ventilation and apnoeas and promoting sympathovagal imbalance, and it is associated with arrhythmias and fatal cardiorespiratory events. In the last few years, opportunities to desensitize hyperactive chemoreceptors have emerged as potential options for treatment of hypertension and heart failure. This review summarizes up to date evidence of chemoreflex physiology/pathophysiology, highlighting the clinical significance of chemoreflex dysfunction, and lists the latest proof of concept studies based on modulation of the chemoreflex as a novel target in cardiovascular diseases.

Graphical Abstract

Clinical conditions associated with chemoreflex overactivity.

Decreased baroreflex sensitivity, increased ergoreflex sensitivity, increased sympathetic drive, renin-angiotensin-aldosterone system (RAAS) activation, phasic hypoxia/reoxygenation and inflammation that occurs in hypertension (HT) and heart failure (HF) promote increased peripheral and central chemosensitivity, leading to adverse clinical consequences, such as ventilatory instability, apnoeas (central apnoeas [CA], high loop gain obstructive apnoeas [OSA_HLG], mixed apnoeas [MA]), increased and unstable blood pressure (BP), increased arrhythmic risk, arousability, dyspnoea and anxiety. Several treatments may act directly or indirectly on the chemoreflex system to reset a physiologic autonomic and respiratory equilibrium, with positive effects on the cardiovascular system.

Acet: acetazolamide; BAT: baroreflex activation therapy; BB: beta-blockers; Busp: buspirone; CBD: carotid body denervation; CPG: central pattern generator; Dyn CO₂: dynamic carbon dioxide administration; Exer: exercise; HCSB: Hunter-Cheyne-Stokes breathing; CVLM: caudal ventrolateral medulla; HF treat: heart failure treatment; Hypn: hypnotics; 5-HT: 5-hydroxytryptamine neurons (serotonergic); Xant: xantines; LC: locus coeruleus; O₂: oxygen administration; ORi: orexin inhibitors; P2XRi: P2X receptor inhibitors; NIV: non invasive ventilation; NTS: nucleus tractus solitarii; PNS: phrenic nerve stimulation; RTN: retrotrapezoid nucleus; RVLM: rostral ventrolateral medulla; SNS: sympathetic nervous system.

1. Introduction

The chemoreflex is a closed loop reflex, which physiologically adjusts ventilation in response to hypoxia or hypercapnia.^1,2 It also matches ventilation with end-organ perfusion by altering sympatho-vagal balance, in concert with high and low pressure baroreflexes and ergoreflex.¹

If the chemoreflex is dysregulated, either because of an increased or decreased activity, cardiorespiratory control misaligns, resulting in hyper- or hypoventilation, periodic breathing and apnoeas, sympathetic nervous system overactivation and life-threatening arrhythmias, hypo- or hypertension,^1,3 and increased likelihood of fatal apnoeas after seizures.⁴

In cardiovascular diseases, the prevalence and prognostic relevance of chemoreflex dysfunction makes it a promising therapeutic target.^3,5 Some novel treatment options are already available for conditions in which the chemoreflex is overactive, such as hypertension, apnoeas and heart failure (HF),^3,5,6 while interventions to stimulate hypoactive chemoreceptors, mainly in neurological diseases,⁷ have not as yet been studied systematically.

This Review will summarize up to date evidence of chemoreflex pathophysiology, highlight the clinical significance of chemoreflex dysfunction, and list the latest proof of concept studies based on chemoreflex modulation as a novel target in cardiovascular diseases.

2. Chemoreflex physiology

A detailed description of chemoreflex physiology has been reviewed elsewere¹ and is beyond the scope of this review. In summary, there are two principal sets of chemoreceptors (CR): peripheral and central CR (Figure 1 and Table 1).

Figure 1. The integrated system of neuroreflexes.

The central chemoreflex (CR) network is a highly integrated system of neurons and glia. Afferents from peripheral CR, baroreceptors, and ergoreflex receptors reach the brainstem at the nucleus tracti solitari (NTS) and rostral ventrolateral medulla (RVLM), where they interact with central CR signals. Together, all those systems regulate ventilation via the central pattern generator (CPG), blood pressure and heart rate via RVLM, and arousal/alertness responses through ascending cortical pathways.

5-HT: 5-hydroxytryptamine neurons (serotonergic); CVLM: caudal ventrolateral medulla; LC: locus coeruleus; SNS: sympathetic nervous system; RTN: retrotrapezoid nucleus.

Table 1.

Peripheral and central chemoreceptors

Peripheral CR	Main neurotrasmitter	Main stimulus increasing firing	Location	Roles and relative contribution of gas response	Role in sleep/wake	Efferents
Type I cells1,9,10	ATP adenosine dopamine acetylcholine	↓ O2 ↓ pH / ↑ CO2 ↓ perfusion ↑ temperature ↑ catecholamines ↑ angiotensin II ↑ lactate ↑ potassium ↓ plasma glucose	Carotid bodies Aortic bodies	100% to O2 ~ 20–30% to CO2	Essential in sleep Important during wakefulness	Carotid sinus branch of the glossopharyngeal nerve → NTS
Type II cells1,9,10	ATP	↑ Purinergic stimulation of P2Y2 receptors	Carotid bodies Aortic bodies	Paracrine sensitization or neural stem cell progenitor of type I cells	Unknown	Type I cells paracrine stimulation or nerve ending of the carotid sinus branch
Central CR	Main neurotransmitter	Main stimulus increasing firing	Location	Roles and relative contribution of central CR to CO2	Role in sleep/wake	Efferents
Serotonergic1,12	5-HT	↓ internal pH	Medullary and midbrain raphé	Physiological pH (7.2–7.4) ↓ ~ 50%	Medullary: Wake, reduced firing during sleep Midbrain: Reduced threshold for REM sleep	Medullary: Pre-BötC, RTN, NTS, phrenic nerve, XII motor nucleus Midbrain: thalamus, limbic system
RTN1,12	Glutamate	↑ CO2/ ↓ pH (↑ stimulation of H+ sensitive channels, TASK2 and GPR4)	Ventral respiratory region, ventral to the VII nerve motor nucleus	Integration center Lower pH (7.0–7.5) Diurnal and nocturnal response ↓ Large proportion of response due to input from other chemoreceptors	Tidal volume: Wake, NREM and REM sleep Respiratory rate: quiet wake and NREM sleep	rVRC, cVRC, pre-BötC, BötC, NTS, Kölliker-Fuse nucleus
Orexin1,12	Orexin	↑ CO2/ ↓ pH	Lateral hypothalamus Prefornical area Dorsomedial hypothalamus	Mainly diurnal response Exercise response Arousal response ↓ 20–50%	Wake only	Medullary raphe, RTN, preBötC, Kölliker-Fuse nucleus, NTS, RVLM, premotor neurons of phrenic and XII nerves, limbic system
Locus Coeruleus1,12	Norepinephrine	↓ internal pH	Floor of the IV ventricle	Mainly diurnal response Alert response to higher CO2 levels Integration center ↓ ~ 60%	Wake	Sleep/Arousal: Basal forebrain, preoptic regions Autonomic regulation: RVLM, IMLC, caudal raphé
Glia1,12	ATP Adenosine	↑ CO2/ ↓ pH (Na+-induced ↑Ca2+ → ATP release → RTN activation via P2 receptors) ↓O2 (TRPA1 channels)	Diffuse but predominantly VRC	Amplification of central chemosensitivity to CO2	-	Predominantly RTN

Abbreviations: 5-HT, 5-hydroxytryptamine; ATP: adenosine triphosphate; BötC, Bötzinger complex; CR: chemoreceptors; CO₂, carbon dioxide; cVRC, ventral respiratory column; GPR4: guanine nucleotide–binding proteins receptors; IMLC, intermediolateral column; NREM: non-rapid eye movement; NT: neurotransmitter; NTS, nucleus tractus solitarius; preBötC, preBötzinger complex; O₂, oxygen; REM: rapid eye movement; RTN, retrotrapezoid nucleus; RVLM, rostral ventrolateral medulla; rVRC, rostral ventral respiratory column; TASK2: TWIK-related acid-sensitive K+ channel; TRPA1: transient receptor potential cation channel, subfamily A, member 1.

Peripheral CR are sited in the carotid bodies (CB) and in the aortic bodies.^8,9 CB are composed of chemoreceptor type I cells surrounded by glial type II cells. The type I cells are polymodal and respond to changes in arterial PO₂, PCO₂, pH, low perfusion, temperature, catecholamines, lactate, angiotensin II, potassium, and plasma glucose.¹⁰ They release multiple inhibitory or stimulatory neurotransmitters (ATP, adenosine, dopamine, and acetylcholine),^2,11 as summarized in Table 1. Peripheral CR contribute up to 30% of ventilatory response to CO₂¹² and are the main O₂ sensors in the body,^13–15 modulating also the autonomic outflow.^1,2,6,16

Central CR are a complex, highly integrated, neuronal and glial network responsible for regulation of the ventilatory response to PCO₂ (~70–80%)¹² and arousal. Different groups of neurons have been proposed as central CR, operating with different CO₂ thresholds at different times of the day, or under different conditions. Prominent subgroups of central CR include serotonergic, glutamatergic, adrenergic and orexinergic neurons, as well as purinergic receptors on glial cells, as summarized in Table 1. Central CR also regulate autonomic output.^1,6 Peripheral and central CRs have different activating threshold for ventilation and adrenergic responses in healthy individuals,⁶ with potential sex related differences.¹⁷ Indeed, compared to men, women show an attenuated ventilatory response to both peripheral and central chemoreflex activation,^17,18 while they show an augmented sympathethic response to central chemoreflex activation.¹⁷ CR stimulation of higher circuits (thalamus, limbic system) also activates arousal and attention/wake promoting pathways with potentially aversive reactions, such as dyspnoea or panic, secondarily affecting sympathetic discharge and ventilation.¹

Peripheral and central CR are intertwined, with hypo-additive, additive, and hyper-additive responses in animals (depending on the species and experimental preparations used),¹⁹ and additive response in humans.¹⁹ CR are also tightly connected with baroreceptor and ergoreceptors (Figure 1). In animals, baroreflex activation inhibits while deactivation augments the peripheral CR ventilatory²⁰ and vasoconstrictor responses.²¹ Once stimulated, the baroreflex seems to exert an inhibitory action on peripheral CR (hypoxia) in humans,²² while constant peripheral CR stimulation (high altitude hypoxia) was shown to cause baroreflex resetting rather than inhibition.²³ Finally, CR and ergoreceptors are coordinated during exercise to match ventilation with increased metabolic CO₂ production and O₂ consumption.²⁴ During mild exercise, peripheral chemosensitivity to O₂/CO₂ significantly increases in healthy subjects²⁵ and this is secondary to increased sympathetic activity, exercise metabolites stimulating peripheral CR, and to the integration with ergoreceptors afferents at medullary level (Figure 1).²⁶ In this context, the potentiation of CB activation by lactate during exercise may facilitate ventilatory compensation and blunt peripheral lactate release during exercise.²⁷

3. Epidemiology of chemoreflex dysfunction

Enhanced chemosensitivity has been described in arterial hypertension and obstructive sleep apnoea (OSA).^6,28–30 The evidence of heightened peripheral chemosensitivity to hypoxia in spontaneously hypertensive rats (SHR)³¹ was later confirmed in young humans with borderline²⁸ or mild hypertension,²⁹ showing an increased blood pressure response to hypoxia.²⁸ In non-hypercapnic OSA, chronic intermittent hypoxia seems to promote peripheral CR sensitization over time,^30,32 causing sympathetic overactivity during wakefulness via cingulate and thalamic regions,³³ with potential effects also on central CR.

The critical role of increased chemosensitivity has also been extensively studied in subject with HF.^34–36 In a population of 110 patients with systolic HF (mean left ventricular ejection fraction, LVEF 31±7%), an increased chemosensitivity to hypoxia, to hypercapnia or both was observed in 12%, 21% and 28%, respectively.³⁴ Similar prevalence rates were observed in HF by Ponikowski³⁵ and Niewinski,³⁷ who found an increased chemosensitivity to hypoxia in 40–45% of patients, and Javaheri³⁶, who found an increased chemosensitivity to hypercapnia in 30% of patients, respectively.

The epidemiology of chemoreflex activation in patients with coronary artery disease is unclear and may partly depend on the treatment, since ticagrelor has been associated with chemoreflex sensitization and central apnoeas (CA) in this setting.^38,39 In ischemic cerebral disease, increased chemosensitivity to hypercapnia has been described in patients with bilateral stroke presenting with CA,⁴⁰ and may be epidemiologically relevant considering a prevalence of CA of 12% in this context.⁴¹

On the other hand, a decreased chemosensitivity has been described in obesity hypoventilation syndrome, the Pickwickian syndrome (hypercapnic OSA), and some neurological conditions, such as congenital central hypoventilation syndrome and sudden unexpected death in epilepsy.^42,43 In hypercapnic OSA, both hypoxic and hypercapnic ventilatory responses are diminished.⁴⁴

4. Pathophysiology of chemoreflex dysfunction

4.1. Arterial hypertension

Increased chemosensitivity is involved in neurogenic and OSA-related hypertension (Table 2 and Graphical Abstract).^{30–32,45,46}

Table 2.

Hyperactive chemoreflex: epidemiology, pathophysiology, clinical presentation and treatment in heart failure and hypertension

Clinical scenario	Epidemiology	Pathophysiological background	Common signs and symptoms and prognosis	Available treatment options
HT (neurogenic) (OSA)	-	Peripheral CR ↑ SNS, RAAS, ATII production Hypoxia/reoxygenation → ↑ ROS ↓ CB shear stress and perfusion CB hypertrophy ↑ ASIC and TASK receptors ↑ P2X3R expression	Clinical picture Dyspnea ↑ Mechanical obstruction ↑ Sympathetic activation Arrhythmias Sleep disruption	Peripheral CR Bilateral denervation → animal models106 → humans with drug-resistant HT107 P2XR antagonists → animal models108 O2 → humans with OSA-induced HT111
		Central CR ↑ Peripheral CR stimulation ↑ Orexinergic neuron firing ↑ ASIC and TASK receptors in 5HT neurons	Adverse prognosis     -	Central CR Almorexant: OR antagonist → animal models109 Suvorexant: OR antagonist → humans with OSA-induced HT110 SSRI → humans with OSA-induced HT111
HF (systolic)	50 subjects with systolic HF35 ↑ peripheral CR: 40% 20 subjects with systolic HF36 ↑ central CR: 30% 110 subjects with systolic HF34 ↑ peripheral CR: 12% ↑ central CR: 21% ↑ peripheral + central CR: 28% 425 subjects with systolic HF105 ↑ peripheral CR: 9% ↑ central CR: 43% ↑ peripheral + central CR: 20%	Pepipheral CR Hypoxia/reoxygenation → ↑ ROS ↓ CB shear stress and perfusion ↑ SNS, RAAS, ATII production ↑ Endothelin	Clinical picture Dyspnea CA/HCSB ↑ Sympathetic activation Arrhythmias Exercise intolerance Sleep disruption	Peripheral CR Uni/Bilateral denervation → animal models130,131 → humans132 O2/CO2 → humans127–129 H2S inhibitor PAG → animal models137 Caffeine, theophylline: adenosine antagonists → humans134–136
		Central CR ↑ Peripheral CR stimulation ↑ SNS, RAAS, ATII production ↑ LA pressure	Adverse prognosis 80 subjects with systolic HF34 ↑ peripheral CR HR 2.8, 95% CI 1.5–5.5, p=0.002 110 subjects with systolic HF104 ↑ peripheral + central CR HR 7.9, 95% CI 2.5–25.0, p<0.001 425 subjects with systolic HF105 ↑ peripheral + central CR HR 2.9, 95% CI 1.3–6.3, p=0.007	Central CR Exercise training → humans148 Yoga → humans147 8-OH-DPAT: 5HT1A agonist → animal models144,145 Buspirone: 5HT1A agonist → animal models143 → humans (BREATH)146

Abbreviations: 5-HT, 5-hydroxytryptamine; 8-OH-DPAT: 8-Hydroxy-2-(di-n-propylamino)tetralin; ASIC: acid-sensing ion channel; ATII: angiotensin II; CA: central apneas; CB: carotid bodies; CR: chemoreceptors; CI: confidence interval; HCSB: Cheyne/Stokes respiration; HCVR: hypercapnic ventilatory response; HF: heart failure; HR: hazard ratio; HT: hypertension; HVR: hypoxic ventilatory response; LA: left atrial; O₂: oxygen; OR: orexin; OSA: obstructive sleep apnea; PAG: DL-Propargylglycine cystathionine γ-lyase inhibitor; P2XR: purinergic receptor 2X; CR: chemoreceptor; RAAS: renin-angiotensin-aldosterone system; ROS: reactive oxygen species; SNS: sympathetic nervous system; SSRI: selective serotonin reuptake inhibitors; TASK: two-pore-domain potassium channel; TCA: tricyclic antidepressants.

Activation of the sympathetic and renin angiotensin aldosterone system (RAAS) seem the primary mechanism,^45,46 while CB chronic hypoperfusion secondary to atherosclerotic processes, and chronic hypoxia/reoxygenation, may play a supplemental role in some phenotypes (i.e., OSA).^45,46 CB hypertrophy and a higher expression of both ASIC and TASK channels, which are responsible for type I cells’ response to pH changes, have also been described.⁴⁷ Finally, increased levels of adenosine have been observed in chronic intermittent hypoxia, as in OSA, increasing peripheral chemosensitivity.⁴⁸

In the long term, persistently increased sympathetic activity – secondary to intermittent hypoxia, stimulation of peripheral CR along with baroreceptor resetting – could promote vascular remodeling and persistent elevation of blood pressure during sleep, but also during wakefulness,⁴⁹ even though other mechanisms may be important as well (i.e., OSA-related arousals or dysbiosis).^50,51

A potential role of central CR must be considered too, especially of the orexin system, which is upregulated in SHR, with increased chemosensitivity to hypercapnia even after hyperoxic blunting of peripheral CR.^45,46 Furthermore, an association between OSA and genetic variants of glial-derived growth factor and the 5-hydroxytryptamine 2_A-receptor (5-HT2_A) has been identified in a large cohort of European and African-American subjects.⁵²

4.2. Heart failure

In HF, both peripheral and central CR are frequently overactive (Table 2 and Graphical Abstract).

Chronic CB hypoperfusion, alternating hypoxia/reoxygenation cycles and reduced shear stress induce local production of reactive oxygen species,⁵³ while persistent activation of sympathetic and RAAS upregulates pro-oxidant and downregulates antioxidant systems, impairing potassium/calcium signaling and cellular excitation.^54,55 Angiotensin II also promotes the secretion of ATP from type II CB cells, causing purinergic postsynaptic sensitization. However, angiotensin receptor antagonists do not appear to impact significantly on peripheral chemosensitivity.^55,56 Chronic hypoxia/reoxygenation also upregulates the endothelin system, with subsequent vasoconstriction further reducing CB perfusion.⁵⁴ Finally, HF patients also exhibit higher plasma levels of adenosine, which can augment peripheral chemosensitivity to O₂, promoting CA and reflexively increasing efferent muscle sympathetic nerve activity (MSNA).^55,57,58

The molecular mechanisms of increased central chemosensitivity in HF are less defined. First, tonic stimulation from peripheral to central CR has been hypothesized.⁵⁹ A role of increased sympathetic drive and RAAS activation has also been postulated, as alpha-2 agonists reduce central chemosensitivity and angiotensin I receptors were identified in the rostral ventrolateral medulla.^60,61 Finally, increased left atrial pressure and pulmonary J receptor stimulation may favor central CR sensitization through brainstem integration.⁶² Indeed, the prevalence of CA increases with sign of elevated atrial pressure in HF, across the whole spectrum of LVEF.⁴⁹

5. Diagnostic assessment of the chemoreflex

There are several ways to assess the chemoreflex in humans, but they all consider the O₂ and CO₂ contributions to ventilatory changes separately.⁶³ Chemoreflex function may be assessed both in terms of sensitivity (i.e., response to gas challenges) or tonic activity (for peripheral CRs).

As for chemoreflex sensitivity, the first test developed were based on steady state methods.⁶⁴ Steady-state tests consisted of maintaining inspired fractional concentrations of O₂ and CO₂ for many minutes (5–20) to allow for equilibration in all tissues.

To overcome variation in cerebral blood flow due to prolonged hypoxia/hypercapnia, transient tests were developed. To study peripheral chemosensitivity to CO₂, a single breath of CO₂ (13%) is given and ventilatory response determined within the first 20 seconds,⁶⁵ to exclude the slower responding central CR. The peripheral chemosensitivity to O₂ is commonly evaluated after two to eight breaths of pure nitrogen, causing a fast drop in O₂ saturation (SaO₂) and a rise in ventilation in about 10 seconds.^66,67

Another way to assess the chemoreflex sensitivity is to use the rebreathing techniques.^68,69 In the rebreathing test for O₂ sensitivity, the subject breathes in a closed circuit so that as PO₂ decreases ventilation rises, accordingly; PCO₂ is kept constant through a scrubbing circuit.⁷⁰ To assess chemosensitivity to hypercapnia, normoxia is maintained by adding O₂ to the circuit.^34,71 The slope of the regression line between SaO₂ and ventilation or between CO₂ and ventilation is a measure of chemosensitivity to hypoxia or hypercapnia, respectively. If the test is performed in hyperoxic conditions, as in the standard Read`s rebreathing technique (5% CO₂ + 95% O₂), central chemosensitivity is assessed (blunting peripheral CRs):⁶³ the addition of CO₂ in the rebreathing bag is done to increase PaCO₂ and etCO₂ rapidly up to the mixed venous PCO_2. Peripheral chemosensitivity can be then estimated by subtraction, starting this time from a normoxic or hypoxic hypercapnic test, for both ventilatory and noradrenergic responses (Figure 2).⁸ If hyperventilation is performed before the rebreathing, as developed by Duffin, the CR recruitment threshold to CO₂ may be identified.⁷²

Figure 2. Rebreathing test.

Hypoxic hypercapnic (filled circles) and hyperoxic hypercapnic (open circles) rebreathing tests displayed plotting end tidal carbon dioxide (P_ETCO₂) as a function of ventilation (A) or muscle sympathetic nervous activity (MSNA) in a healthy volunteer. Big circles refer to bin-averaged values (from different subjects) of ventilation (left panel) and MSNA (right panel) for 2 mmHg intervals of P_ETCO₂ variation. Whiskers refer to the standard deviation of P_ETCO₂ (horizontal whiskers), and either ventilation or MSNA (vertical whiskers). Fitted double linear models are superimposed over the data (hyperoxia, black and hypoxia, grey). Gently taken and adapted (with permission) from Keir et al.¹³

On the other hand, the tonic activity of peripheral CRs has been shown to contribute to both resting ventilation and hemodynamic control in humans, by interacting with baroreflex function.^73–75 Notably, such contributions may be easily estimated by inhibiting peripheral CR, through either hyperoxia or low-dose dopamine infusion, and assessing the subsequent changes in minute ventilation, heart rate, and blood pressure.^73–75

6. Clinical significance of chemoreflex dysfunction

6.1. Hypertension and obstructive sleep apnoea

Potentiated chemoreflex-mediated sympathetic vasoconstriction has been documented in patients with neurogenic hypertension, especially in patients with OSA,⁷⁶ and it is reversed by hyperoxia, which blunts peripheral CR.⁷⁷ In the so-called high loop-gain OSA and mixed apnoeas, enhanced chemoreflex may also cause ventilatory overshoot after apnoea, with hypocapnia causing hypotonia of the upper airway musculature^78,79 and subsequent hypopnea and/or apnoea, by lowering PaCO₂ towards or below apneic threshold.

OSA have been associated both with brady- and tachyarrhythmias, especially during sleep, and a role for CR has been suggested for both phenomena.⁸⁰ While CR stimulation in the absence of ventilation (late apnoea) may cause vagal reflex inhibition of the heart favoring sinus arrest or atrioventricular blocks,^80,81 OSA-induced CR oversensitization during normal breathing may promote ectopics, atrial fibrillation, and ventricular arrhythmic events, also during wakefulness.^80,81

6.2. Heart failure

In HF, increased chemoreflex sensitivity, alongside increased plant gain (i.e., increased changes in PCO₂ for a given change in ventilation) and prolonged lung-to-chemoreceptor circulation time promote ventilatory instability, as hypothesized by mathematical models,⁸² and confirmed in both animal^54,55,59 and human studies.^34–36 Indeed, patients with HF often exhibit a respiratory pattern characterized by alternating phases of hyperventilation and CA, named Hunter-Cheyne-Stokes breathing (HCSB).^34–36

Notably, these 3 mechanisms are state independent (i.e., present while awake and asleep), and consequently CA/HCSB could be observed during wakefulness at rest and exercise.^83,84 Nevertheless, CA/HCSB is usually most pronounced during non-rapid eye movement (NREM) sleep, due to predominance of the chemical control of breathing and the unmasking of the apneic threshold.⁸⁵ Increased arterial circulation time may prolong CA/HCSB cycle, especially in patients with reduced LVEF, as a result of delayed transfer of information regarding the level of gas tensions in the pulmonary capillary bed to chemoreceptors.^58,86–89

Increased chemosensitivity has been associated with more severe dyspnoea, as expressed by a higher NYHA class, in HF patients.²⁵ Indeed, CR firing increases the respiratory motor output that can be consciously felt as unpleasant, and signals with ascending connections to areas of the limbic system and cortex involved with the perception of dyspnoea,⁹⁰ being together with CB lactate sensing²⁷ and increased ergoreflex sensitivity during effort a main determinant of HF symptoms.

HF patients with elevated chemosensitivity also present exercise limitation with lower peak O₂ consumption and ventilatory efficiency (increased VE/VCO₂ slope),^25,34,91 which correlate with CA severity in HF.⁹² Notably, chemoreflex downregulation with hyperoxia,⁹³ or dihydrocodeine administration⁹⁴ improves exercise tolerance, ventilatory efficiency, and dyspnoea in HF patients. No study has yet investigated the relationship between overactive CR and exertional oscillatory ventilation, another ominous prognostic marker in HF, so far mainly related to the haemodynamic compromise during exercise.⁹⁵

Furthermore, patients with overactive CR show signs of sympathovagal imbalance with sympathetic predominance, and a higher risk of atrial and ventricular arrhythmias.^{34,35,96–98} In animal models, peripheral CR overactivity has been linked with increased sympathetic drive to the heart,⁹⁹ leading to adverse remodeling and proarrhythmic events, and to the kidneys,¹⁰⁰ causing reduced renal blood flow and predisposing to cardiorenal syndrome. Chemoreflex has also been linked to pulmonary vasoconstriction during CA and right ventricular overload in HF patients.^101,102 In this respect, microneurographic tecniques may further clarify the relationship between chemoreflex overactivation and sympathovagal imbalance in HF.^8,103

Moreover, increased chemosensitivity carries itself a poor prognosis in HF. In patients with systolic HF (n=80, LVEF <45%), recruited in the pre-beta-blocker era, increased peripheral chemosensitivity to hypoxia (cutpoint: >0.72 L/min/%SaO₂, transient hypoxic test) was shown to be an independent predictor of overall death at multivariable analysis.¹⁰⁴

These results were later replicated by Giannoni et al. in two different studies on HF patients.^34,105 In the largest study conducted so far (n=425) in patients with HF (LVEF <50%) on modern treatment, a combined increased chemosensitivity to hypoxia and hypercapnia was found to be an independent predictor of the primary outcome (a composite of cardiac death, appropriate implantable defibrillator shocks and HF hospitalization) and the secondary outcomes (i.e., each individual component of the primary outcome). A particulary detrimental outcome was observed in patients with both increased chemosensitivity and decreased baroreflex sensitivity (Figure 3). Notably, adding chemoreflex and baroreflex sensitivity to a multivariate prognostic model (including age, aetiology, LVEF, NT-proBNP, renal function, peak O₂ consumption, and VE/VCO₂ slope) significantly improved risk prediction.¹⁰⁵

Figure 3. Abnormal baroreflex (BRS) and chemoreflex sensitivities (CRS) in patients with chronic heart failure (HF).

Abnormal BRS and CRS may be frequently observed in chronic HF patients on modern therapies and both contribute to autonomic dysfunction. Abormal BRS is also associated with functional impairment and a significantly higher risk of cardiac death, while abnormal CRS (particularly when both the hypoxic [HVR] and the hypercapnic ventilatory responses [HCVR] are heightened) is associated with ventilatory instability and higher risk of HF hospitalization. When both reflexes are abnormal, the risk of adverse events is very high, with less than 10% of patients being free of events over a 60-month follow-up. Taken and adapted with permission from Giannoni A et al.⁹²

7. Therapeutic approaches to chemoreflex dysfunction

Considering its clinical significance, chemoreflex has been recently proposed as a potential therapeutic target (Figure 4).⁵ The optimal criteria to select the patients who could benefit more from CR modulation is still a matter of debate. While CR sensitivity, assessed through ventilatory response, has been extensively associated with poor outcomes,^31,62,101 the study of CR tonicity may also predict clinical response to treatment.¹⁰⁶ Furthermore, the possible role of either clinical or biohumoral markers as surrogates of CR function (not routinely assessed) to select patients for chemoreflex modulation remains to be clarified.¹⁰⁷

Figure 4. Potential treatments of chemoreflex hyperactivation in heart failure and hypertension.

In hypertension, surgical denervation of carotid bodies or their silencing with oxygen (O₂) has also been proposed to treat hyperactive peripheral chemoreceptors (CR), together with pharmacological modulation of the purinergic signaling with the novel P2X receptor (P2XR) antagonists. Similarly, pharmacological treatment of hyperactive central CR has also been tested with orexinergic receptor (OR) antagonists (almorexant and suvorexant) in neurogenic hypertension and high loop-gain obstructive apnoeas. In heart failure, modulation of carotid bodies with O₂ and CO₂ (especially given dynamically during CO₂ drops following hyperventilation), with caffeine or theophylline acting on the adenosine pathway, or with surgical denervation have been attempted to target hyperactive peripheral CR. Central CR modulation has also been attempted, both with exercise training and yoga and with pharmacological agents as buspirone, by decreasing the serotonergic firing.

Chemoreflex modulation has been attempted in hypertension, mainly in preclinical models. In SHR, bilateral, but not unilateral, CB denervation increased baroreflex sensitivity (+32%) and decreased renal sympathetic nerve activity (−56%), low frequency component of heart rate variability (−28%) and arterial pressure by 17 mmHg (−12%).¹⁰⁸ In a following safety and feasibility trial in humans, unilateral CB denervation was effective only in those with increased chemosensitivity (53% of the total population), though some feedback resetting occurred over time.¹⁰⁹

A promising alternative is the pharmacological modulation of chemoreflex. SHR overexpress purinergic P2X3 receptors and show increased peripheral chemosensitivity,¹¹⁰ which was decreased by either the local or systemic delivery of two P2X3 receptor antagonists, AF-533 and AF-219, blunting adrenergic outflow and blood pressure.¹¹⁰ Although the possibility of modulating central CR has been tested in mice with neurogenic hypertension with promising results with an orexin antagonist, almorexant,¹¹¹ its use was discontinued for safety concerns in humans. Suvorexant showed instead neutral results in patients with OSA.¹¹²

The subgroup of patients with HF/hypertension and OSA seems the most complex to manage,¹¹³ since chemosensitivity may be either increased (high loop-gain) or decreased (low loop-gain). Indeed, downregulating peripheral CR with O₂ decreased the apnoea burden by 53% in the former group, but only by 4% in latter. On the contrary, stimulation of noradrenergic and serotonergic CR with selective serotonin reuptake inhibitors, such as fluoxetine and paroxetine, was partly effective in low loop-gain OSA during NREM but not during REM sleep.¹¹³ In hypercapnic OSA progesterone administration has been also proposed.^114,115

In HF, the possibility to modulate the hyperactive CR is particularly appealing in light of the prognostic role of chemoreflex in this setting.^34,104 The phenotypic therapy of both OSA and CA in HF has been extensively reviewed elsewhere.¹¹⁶

Beta-blockers reduced peripheral chemosensitivity to hypoxia (carvedilol and nebivolol) and the central chemosensitivity to hypercapnia (carvedilol),¹¹⁷ with positive effect on ventilatory efficiency during exercise.¹¹⁸ RAAS antagonists were shown to normalize both sympathetic and ventilatory response to hypoxia in HF rabbits.¹¹⁹ Finally, cardiac resynchronization therapy reduced chemosensitivity to hypercapnia (by around 22%) 4–6 months after implantation.¹²⁰ Left-ventricular-assist devices, on the other hand, only improved ventilatory efficiency during exercise with increased pump speed.^121,122 Finally, heart transplant recipients showed a persistently higher peripheral ventilatory and sympathetic response to hypoxia, but not to hyperoxic hypercapnia, 7±1 years after surgery compared to controls.¹²³

Approaches acting on ventilation rather than on respiratory gases, such as continuous positive airway pressure (CPAP)¹²⁴ and phrenic nerve stimulation,¹²⁵ may favorably decrease both oscillations in respiratory gases and, secondarily, chemoreflex stimulation.¹²⁶ While CPAP was shown to decrease the chemoreflex sensitivity to hypoxia but not to hypercapnia in patients with OSA at least in the short term,^126,127 the effects of CPAP and phrenic nerve stimulation on chemosensitivity in patients with HF and CA are still unknown. On the other hand, acetazolamide, a drug that reduces CA in HF likely by decreasing plant gain,¹¹⁶ also blunts the peripheral chemosensitivity to hypoxia, while increasing both chemosensitivity to hypercapnia and VE/VCO₂ slope during exercise.¹²⁸

CO₂ enriched gas (4%) was able to abolish CA/HCSB in HF patients,¹²⁹ even though it was later demonstrated that CO₂ also promotes CR stimulation with sleep disruption and sympathetic overactivity,¹³⁰ unless delivered by automatic approaches minimizing the dosage.¹³¹ On the other hand, a constant O₂ flow (1–5 L/min) reduced peripheral CR activity and also nighttime CA by ~50%, improving exercise performance and decreasing sympathetic overactivity.¹³⁰ A phase 3 randomized clinical trial (LOFT-HF, NCT03745898) on the prognostic effects of nocturnal O₂ therapy in patients with stable HF and CA was eary terminated for delayed site activation and low recruitment during the COVID-19 pandemic.

In animal models of HF obtained by pacing overdrive (rabbit) or coronary artery ligation (rats), bilateral CB denervation stabilized breathing, reduced sympathetic outflow to the heart and kidney and restored baroreflex sensitivity.^132,133 This translated into decreased risk of arrhythmias, positive reverse left ventricular remodeling, and improved survival.¹³³ A similar approach was attempted in 10 HF patients with increased peripheral chemosensitivity.¹³⁴ CB surgical resection was performed by a lateral approach to the carotid bifurcation, with CB macroscopically identified and histologically confirmed. After unilateral (n=4) or bilateral (n=6) CB denervation, peripheral chemosensitivity decreased by 70%, mainly in patients with bilateral resection. This led to a decrease in sympathetic activity, VE/VCO₂ slope and increased exercise duration, with no changes in natriuretic peptides, LVEF, or quality of life indexes. However, decreased SaO₂ at night was observed in 60% of patients, with one case of worsening OSA needing ventilatory support and two deaths during follow-up.¹³⁴ A unilateral approach or, alternatively, a reversible pharmacological CB modulation may hence be preferred, considering the key role of CB in hypoxia sensing.¹³⁵ In this setting, intra-carotid injection of adenosine has been proposed to assess the residual chemoreflex function after carotid body ablation.¹³⁶ From a single study run in 6 male CHF patients, it seems that after CB denervation ventilatory and blood pressure responses to hypoxia are strongly reduced, while heart rate response (tachycardia) is preserved, suggesting a potential role for other hypoxia sensors as aortic bodies also in humans.¹³⁷

Some drugs potentially acting on peripheral CR have been also tested. Caffeine, an adenosine receptor antagonist, decreased MSNA response to handgrip maneuver in HF¹³⁸ and peripheral CR activity in an animal models,¹³⁹ while theophylline, another adenosine antagonist, halved CA and improved SaO₂ in HF patients,¹⁴⁰ even though this effect might primarly be driven by a reduction in the plant gain,¹⁴¹ similarly to acetazolamide.¹¹⁶

Furthermore, a hydrogen sulfide inhibitor, DL-Propargylglycine cystathionine γ-lyase inhibitor, was shown to decrease the peripheral chemosensitivity stabilizing breathing and ameliorating autonomic indexes in a rat model of HF.¹⁴² Finally, in an animal model of post-infarction HF, simvastatin treatment significantly improved respiratory variability through the upregulation of Krüppel-like factor 2 and endothelial nitric oxide synthase, whose reduced expression within the CB and the nucleus tracti solitari is associated with enhanced chemosensitivity^143,144.

Another interesting possibility is to target the central CR. In particular, the serotonergic system seems a promising target, since it contributes to 30–50% of the chemosensitivity to hypercapnia in the physiological range and can be modulated by well-characterized medications. Buspirone, a presynaptic 5-HT1_A receptor agonist long and safely used in general anxiety disorder, was shown to decrease the chemosensitivity to hypercapnia and stabilize breathing in a mouse model of hypoxia-induced apnoeas.^142,145 Similar results were obtained with another 5-HT1_A receptor agonist, 8-hydroxy-2-(di-n-propylamino)tetralin, in rats and piglets.^146,147 A recent double-blind randomized placebo-controlled crossover trial, the BREATH study,¹⁴⁸ has investigated the effect of oral buspirone administration (45 mg) in patients with systolic HF (n=16). After 1-week of treatment, buspirone decreased central chemosensitivity to hypercapnia and reduced CA throughout the 24-hour period compared to placebo, without significant side effects.

Lastly, modulation of respiration using slow breathing as in yoga,¹⁴⁹ exercise physical training,¹⁵⁰ or baroreflex stimulation^151,152 may be successful strategies in decreasing chemosensitivity in HF and hypertension by acting on brainstem integration circuits.

8. Conclusions

The chemoreflex is a closed loop reflex, tightly integrated with baroreflex and ergoreflex, that matches ventilation and sympathetic drive with blood gases. Its dysfunction contributes to highly prevalent pathologies such as hypertension and HF.

A deeper understanding of the pathophysiological basis of chemoreflex hyper- or hypoactivity, coupled with the recent insight into respiratory and sympathetic chemoreflex responses, has led to the development of interesting new therapeutic approaches both in hypertension and HF, opening an exciting new research field and prompting the design of larger and longer clinical studies with particular relevance towards the selection of patients that would benefit the most from chemoreflex modulation strategies.

Table 3.

Hypoactive chemoreflex: epidemiology, pathophysiology, clinical presentation and treatment in congenital central hypoventilation syndrome and sudden unexpected death in epilepsy.

Hypoactive Chemoreflex
Clinical scenario	Epidemiology	Pathophysiological background	Common signs and symptoms	Available treatment options
CCHS	-	Peripheral CR Mutations in Phox2b gene → polyalanine residues expansion → frameshift mutations     Central CR Mutations in Phox2b gene → polyalanine residues expansion → frameshift mutations	Breathing abnormalities Impaired CO2 arousability Apneas 24-h ventilatory support ↑ Sympathetic activation	Central CR Progestins112
SUDEP	-	Peripheral CR     -     Central CR ↓ staining of TH in raphe nuclei ↓ 5-HT innervation of RTN ↓ substance P in respiratory areas	Post-ictal apnea with respiratory and then cardiac arrest Death Impaired CO2 arousability	Central CR SSRI (fluoxetine, paroxetine)7 Ketogenic diet7

Abbreviations: H₂S: hydrogen sulfide; 5-HT, 5-hydroxytryptamine; CR: central chemoreceptors; CCHS: congenital central hypoventilation syndrome; O₂: oxygen; PCR: peripheral chemoreceptor; PAG: Phox2b: paired-like homeobox 2b; RTN: retrotrapezoid nucleus; SSRI: selective serotonin reuptake inhibitors; SUDEP: sudden unexpected death in epilepsy; TH: tryptophan hydroxylase.

List of abbreviations

CA

central apnoeas

CB

carotid bodies

CPAP

continuous positive airway pressure

CR

chemoreceptors

HCSB

Hunter-Cheyne-Stokes Breathing

HF

heart failure

5-HT

5-hydroxytryptamine

LVEF

left ventricular ejection fraction

MSNA

muscle sympathetic nerve activity

NREM

non-rapid eye movement

OSA

obstructive sleep apnoeas

RAAS

renin-angiotensin-aldosterone system

REM

rapid eye movement

SaO₂

oxygen saturation

SHR

spontaneously hypertensive rats