Contrary to our expectations naloxone had no effect on any of the investigated variables, neither in normoxia nor in hypoxia. It follows that, at least in trained healthy young male subjects, during incremental exercise tests, in normoxia and normobaric hypoxia with an Fi O2 of 10.65 (equivalent to approximately 5,000 m), endogenous opioid receptors are not involved in the ventilatory and heart rate responses to exercise nor in the sensation of overall levels of perceived exertion or those pertaining to breathing or cycling effort specifically.
Could it be that there was insufficient blockade of opioid receptors? This seems unlikely. The ‘normal’ dose for clinical use of naloxone is 1–4 mg, largely sufficient for full reversal of the effects of exogenously administered opioids and to trigger withdrawal symptoms . Naloxone hydrochloride is partly actively transported through the blood–brain barrier and reaches higher central nervous system (CNS) concentrations than in the plasma . Positron emission tomography studies showed that with 1 mg naloxone, 50% of opioid receptors in the CNS were blocked . Santiago and Edelman  recommended a minimal dose of 0.1 mg/kg for peripheral and central receptor blockade. We used 30 mg, i.e. 0.40 ± 0.04 mg/kg, a dose that is four times in excess, to compare our results to those of a previous study . Naloxone has a half-life of about 1 h . Our subjects started exercising within 5 min after injection and reached exhaustion in less than 30 min, largely inside the therapeutic time window. Finally, even though admittedly anecdotal, several subjects indicated missing their habitual post-exercise ‘high’ and feeling somewhat ‘bland’ instead, which after the breaking of the randomization key appeared to correspond to the naloxone experiments, suggestive of naloxone blocking the effect of endogenous opioid release post-exercise, possibly related to the ‘feeling good’ after a workout [20, 21].
Opioids and dyspnoea
What can explain the difference between our findings and the repeated finding of a role for opioids in patients with dyspnoea? For example, naloxone increased dyspnoea sensation in asthmatics when challenged with metacholine, and similar findings were reported for chronic obstructive pulmonary disease (COPD) patients during exercise [22, 23]. Jensen et al.  compared inhaled nebulized fentanyl citrate (an opioid analogue) to placebo in COPD patients and found dyspnoea attenuation and improved exercise performance. Both sensations of intensity and unpleasantness were affected. There are similar observations in patients with terminal cancer-related dyspnoea , and opioids have a role as therapeutic means in dealing with dyspnoeic patients in general [5, 6, 25].
One explanation for an absence of effect in our study may reside in the complexity of what is covered by the term dyspnoea. The American Thoracic Society defined dyspnoea as ‘a subjective experience of breathing discomfort that consists of qualitatively distinct sensations that vary in intensity’ [5, 26]. Dyspnoea includes both sensory (intensity) and affective (unpleasantness) components . Distinct mechanisms and afferent pathways are associated with different sensory qualities (notably work/effort, tightness and air hunger/unsatisfied inspiration); distinct sensations most often do not occur in isolation, and dyspnoea sensations also vary in their unpleasantness and in their emotional and behavioural significance . In patients, the affective component is often of great importance , whereas it is of less significance in a healthy athletic subject during a non-threatening challenge like an exercise test limited in time. A patient with COPD associates the difficulty to breathe during an exercise test with fear to suffocate, whereas a healthy subject will not . Nevertheless, in healthy subjects, intravenous morphine sulphate reduces the discomfort (‘air hunger’) induced by hypercapnia to a similar extent as observed during clinical studies .
Breathing during heavy exercise in severe hypoxia is accompanied by dyspnoea with affective aspects . In the present study, the subjects quantified the rate of perceived exertion in the classic sense originally proposed by Borg , i.e. the central generated sensation of effort . Affective components of unpleasant sensations related to the exercise like leg pain or air hunger were not explicitly quantified. There are two different dimensions of exercise-related sensations: (1) one reflecting the sensation of effort, likely centrally generated and unrelated to afferent feedback and (2) one related to afferent feedback from various tissues involved in the effort . Aliverti et al.  reported that breathing RPE was uniquely related to total respiratory power output at low and high altitudes (hypobaric hypoxia, 4,559 m). In the present study, in conditions of normobaric hypoxia, the tendency for a leftward shift of the curve relating breathing RPE to V'E (see Figure 1) suggests that breathing RPE may have been influenced by other parameters but that the endogenous opioid system was not involved. However, even if an affective component of the sensation of breathing was influenced by hypoxia and/or naloxone in our study, this did not have an effect on performance.
Ventilatory response to exercise
Anatomical and pharmacological evidence suggests that endogenous opioids play a role in the control of breathing. Mu (μ), delta (δ) and kappa (κ) opioid receptors are present in brainstem areas involved in respiration, and endogenous opioids like endorphins, enkephalins, dynorphins and endomorphins are found in medullary and pontine respiratory regions [6, 31]. The depressant effects of exogenous opioids on ventilation are well known, and also, endogenous opioids are thought to be tonically active and have a depressant effect on ventilation . Exogenous opioids also have a strong depressant effect on the hypoxic ventilatory response in animals and humans . By contrast, on a local CNS level, endogenous opioids may exert an excitatory modulation of hypoxia-induced hyperventilation by acting on μ-receptors in the rostral medullary raphe at least in a rat model , illustrating that the overall systemic effects of exogenous opioids are not necessarily indicative of the role of endogenous opioids at specific sites in the CNS. Akiyama et al.  injected healthy subjects with naloxone hydrochloride and found an increase of both ventilatory (V'E) and mouth pressure (Pm) responses to hypoxic progressive hypercapnia with inspiratory flow-resistive loading. Ward and Nitti  injected the opioid agonist sufentanil in trained athletes during exercise and found that the ventilatory response to exercise was reduced. We expected that the injection of 30 mg of naloxone would increase the ventilatory response to exercise. Our results indicate no effect of a general blockade of opioid receptors on the ventilatory response to incremental exercise in normoxia and hypoxia. It follows that it is unlikely that endogenous opioids play any important role in the ventilatory response to an incremental exercise challenge in normoxia and its increase in hypoxia, at least in healthy, young trained male subjects.
Humans frequently report positive feelings during and after endurance efforts (‘runner's high’) that include both central effects (improved affect, sense of well-being, anxiety reduction, post-exercise calm) and peripheral effects (reduced pain sensation) . Imaging studies provided evidence of endogenous opioid release in fronto-limbic brain regions after physical exercise correlated to perceived euphoria . Endocannabinoids are also thought to play a role in runner's high . Paulev et al.  compared naloxone (0.8 mg i.v.) to placebo during a Cooper test (running the longest possible distance within 12 min) in trained subjects. Performance was not influenced by naloxone, but perception of muscle pain was enhanced with naloxone. Sgherza et al.  compared incremental exercise capacity in 18 subjects under naloxone and placebo and reported a significant reduction with naloxone. They concluded that after naloxone administration, in laboratory conditions, endurance trained subjects stop exercise at lower levels than those under placebo, suggesting that peak exercise capacity was limited by the individual's perception of exertion, exacerbated by a lack of effect from endogenous opioids, rather than by physiological fatigue. Even though several subjects in our study reported feeling ‘bland’ after exercise under naloxone, we did not see the reduction in exercise capacity reported by Sgherza et al.  in normoxia. Their results were based on 18 subjects, and earlier studies with fewer subjects and lower dosage of naloxone had failed to find significant effects. We therefore cannot exclude that our results, obtained in ten subjects, are limited by type-II error, but given the absence of any change in the variables monitored, it seems quite unlikely that in hypoxia an effect of opioid blockade plays any important role.
Apart from their involvement in noxious signalling, spinal cord level opioid receptors may play a role in the exercise pressor response. Amann et al. [38–40] used lumbar intrathecal fentanyl to block central projection of μ-opioid-receptor-sensitive group III/IV muscle afferents from the lower limbs during different exercises: single-leg mild to moderate intensity knee extensor exercise, 5-km cycling time trial exercise, incremental cycling exercise and constant load exercise to exhaustion. Afferent blockade had no effect on central and peripheral haemodynamics or ventilation at rest, but during exercise cardiac output, mean arterial pressure and femoral blood flow were attenuated in the knee extensor exercise, pacing during the time trial was severely perturbed, and heart rate, blood pressure and ventilation were impeded during cycling at 80% of aerobic maximum. These findings suggest an important role of group III/IV muscle afferents in the regulation of the cardiorespiratory response to rhythmic exercise. In rats, Tsuchimochi et al.  found that stimulation of peripheral μ-opioid receptors attenuated the exercise pressor reflex and that this effect could be blocked with naloxone. These findings in humans and animals would suggest that endogenous opioids play a role in the regulation of muscle afferent traffic, but from our results, it appears that in healthy, young, physically active subjects doing a 20- to 30-min incremental exercise challenge in normoxia or severe hypoxia, endogenous opioids are not attenuating the normal cardiovascular response to exercise through the stimulation of type III/IV afferent μ-receptors at the spinal level, leaving the question on their role and significance open.
Acute normoxia switch
The limitation of exercise performance in severe hypoxia is still not well understood [11–13]. Lack of oxygen is the reason, but it remains unclear what mechanism in hypoxia leads to an earlier disengagement from an exercise challenge as compared to normoxia. Verges et al.  argue that because biochemical, electromyographic and mechanical signs of muscle fatigue at exhaustion are reduced in severe hypoxia compared with normoxia, muscle metabolic fatigue is not the main factor responsible for impaired whole body exercise performance, as proposed before [12, 15]. Our subjects were able to continue cycling after reaching their maximum in hypoxia when switched to normoxia. iEMG, an indicator of locomotor muscle drive, increased upon the normoxia switch, a finding arguing in favour of supraspinal limitation of exercise performance in hypoxia by a rapidly reversible withdrawal of motor drive. Amann et al.  found that the degree of hypoxia influences the relative role of muscle fatigue in the cessation of dynamic exercise with large muscle groups. Those authors proposed a threshold of SaO2 for a switch from a predominant effect of peripheral fatigue to a predominant effect of CNS hypoxia on central motor output and exercise performance. They found similar levels of muscle fatigue in task failure from constant load exercise to exhaustion when SaO2 averaged 94%, 82% or 76% (Fi O2 0.21–0.12), but not at a SaO2 of 67% (Fi O2 0.10). They suggest the dominance of CNS hypoxia over peripheral muscle fatigue in influencing central motor output below SaO2 levels of 70%–75%. Our results of early exhaustion in hypoxia with SaO2 below 70% and rapid restoration of exercise capacity upon the normoxia switch are in accordance with those contentions.
Limits of performance in severe hypoxia
The origins of the signals leading to the cessation of the central motor drive at exhaustion during heavy large muscle group exercise in severe hypoxia thus remain unclear. Possible candidates controlling the signal to stop exercise include arterial oxygen (O2) desaturation with exercise causing marked central nervous system hypoxia, other factors acting on the respiratory and/or other higher nervous centres, with or without contribution of fatigued respiratory muscles, or the effects of pulmonary hypertension and right ventricular overload [4, 12, 43, 44]. Millet et al.  showed in a biceps brachii repeated isometric contraction model that in pronounced hypoxia (9% O2, SaO2 75%), central drive is diminished independently of afferent feedback and peripheral fatigue and concluded that submaximal performance in severe hypoxia is related directly to brain oxygenation, results corroborating those of Goodall et al.  who also showed that peripheral mechanisms of fatigue contribute relatively more to the reduction in force-generating capacity of the knee extensors following submaximal intermittent isometric contractions in normoxia and mild to moderate hypoxia, whereas supraspinal fatigue plays a greater role in more pronounced degrees of hypoxia.
Our results pertain to a limited sample size of healthy, young, trained male subjects and cannot be generalised. Further limitations to our study design are acknowledged. Prolonged hypoxia leads to ventilatory acclimatisation changing ventilation at rest and during exercise. We performed our experiments in acute normobaric hypoxia, and it is possible that in chronic hypobaric hypoxia, findings would differ. It therefore remains to be described what the effects of opioids or opioid receptor antagonists on breathing sensation and exercise performance in conditions of prolonged hypobaric hypoxia would be. Finally, we used a short incremental exercise protocol to exhaustion. It remains an open question if more prolonged submaximal time trial like exercise would be influenced by opioid receptor blockade.