INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Obstructive sleep apnea (OSA) is characterized by repetitive partial or complete upper airway occlusions (1) that cause a decrease in or cessation of breathing, repetitive decreases in blood oxygen saturation, and sleep fragmentation. Defined by the occurrence of more than 15 decreases or cessations of breathing per hour of sleep (2), OSA affects 2 to 4% of the middle-aged population (3). The disease is now recognized as a major public health problem because patients with OSA have increased mortality (4) and cardiovascular morbidity, including systemic hypertension, atherosclerotic heart disease, and cerebrovascular stroke (5). Neuropsychologic impairment, including excessive daytime sleepiness, is another common feature of the condition (6). Both recurrent hypoxic episodes and arousal-related, end-apneic hyperadrenergic reactions may be responsible for OSA-associated complications. However, few tissue markers have been described that allow the objective quantification of hypoxic tissue damage in OSA. Autonomic nervous system dysfunction has been reported both in patients with chronic airflow obstruction and chronic hypoxemia (7) and in those with intermittent nocturnal hypoxia during OSA (8). Furthermore, in apneic patients with only intermittent hypoxia, there is a right shift of the oxyhemoglobin dissociation curve associated with increased circulating levels of 2,3-diphosphoglycerate (9) and an increased nocturnal urinary uric acid/creatinine ratio, both of which reflect tissue hypoxia (10). On the other hand, erythropoietin production is not increased in OSA (11).

Changes occur in peripheral nerves subjected to hypoxemia resulting from a low blood O₂ concentration such as in chronic obstructive pulmonary disease (COPD), or to damage to the vasa nervorum causing ischemia, such as in diabetes. A modified tolerance of peripheral nerves to transient, experimentally induced limb ischemia has been reported in hypoxic COPD and diabetic patients (12, 13). This tolerance is characterized by abnormal persisting nerve conduction during ischemia. This resistance to ischemic conduction failure (RICF) seems to be the earliest abnormality of peripheral nerve function observed in diabetic patients who go on to develop an obvious neuropathy (14). The tolerance to ischemia may be an adaptative mechanism, since some electrophysiologic abnormalities have been reversed in hypoxic patients by improvement of oxygenation (12), and in diabetic rats by supplemental oxygen inhalation (15).

Chronic hypoxia is a well known cause of peripheral neuropathy (16). However, to our knowledge, no study has been done of peripheral neuropathy related to intermittent hypoxia except for the pharyngeal thermal hypoesthesia described in patients with OSA, which may be related to a pharyngeal neuropathy (19). The aim of the present study was therefore to assess peripheral nerve function with an ischemia test in OSA patients without a recognized cause of neuropathy. This type of dysfunction may represent an early and quantifiable tissue marker of the consequences of nocturnal hypoxemia.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

The study was a case-control study of 10 control subjects and 17 patients with severe OSA, who were selected prospectively according to the following inclusion criteria.

Patient Selection

The 17 patients with severe OSA had an apnea + hypopnea index (AHI)  40 events/h and minimum Sa_O2 < 80%, and were without daytime hypoxemia (Pa_O2 > 65 mm Hg). The 10 age-matched control subjects had no clinical signs or symptoms of OSA or neuromuscular disease. No subject had a recognized cause of neuropathy (diabetes, alcoholism, renal or respiratory failure, malnutrition, known cancer, infections, systemic disease, or use of neurotoxic drugs). Venous blood was taken for estimation of fasting blood glucose, creatinine level, and hepatic enzymes. The study was approved by the University Hospital ethical committee of Grenoble, and was conducted in the sleep and electromyography laboratories of the University Hospital of Grenoble. Informed consent was obtained from all subjects.

Investigations

In addition to a standard neurologic history and examination, including the Epworth sleepiness scale, controls had an overnight oximetry recording done with a transcutaneous finger probe (Biox 3740 Pulse Oximeter; Ohmeda, Louisville, CO) and OSA patients underwent polysomnography (Respisomnographe; SEFAM-Nellcor Puritan-Bennett, Nancy, France). Electrophysiologic studies were done with a Viking II electromyograph (Nicolet Biomedical Instruments, Madison WI), and consisted of tests of tolerance to ischemia.

Sleep Studies

Overnight polysomnography was performed in a standard manner and was scored manually according to Rechtschaffen and Kales' criteria (20). Microarousals were scored according to criteria of the American Sleep Disorders Association (ASDA) (21). Episodes of apnea were defined as complete cessation of airflow  10 s, and hypopnea was defined as a 50% decrease in oronasal airflow for at least 10 s. Apnea/hypopnea events were classified as obstructive, mixed, or central, according to presence or absence of breathing efforts and the AHI (number of episodes of apnea + hypopnea per hour of sleep) was calculated.

Ischemia Test

The ischemia test was performed with surface electrodes, with the subject lying comfortably in the supine position in a warm room, with the abducted right upper limb resting on a soft pillow (Figure 1). The median nerve was supramaximally stimulated at the wrist with square waves of 200-µs duration, delivered at 3 Hz. Mixed nerve action potential was measured with an electrode placed over the median nerve at the elbow, and sensory nerve action potential was recorded through an antidromic technique on digit III. Arm circumference was measured with a measuring tape at the site of the recording electrode used for measurement of mixed nerve conduction. The ground electrode was placed over the right forearm. Ischemia was induced by rapidy inflating the 13-cm-wide cuff of a sphygmomanometer previously placed round the arm at a distance of 5 to 6 cm above the elbow to a pressure of 50 mm Hg above the subject's systolic blood pressure. Recordings were made before ischemia, every 2 min during a 30-min period of ischemia, and during a 6-min period of recovery. Skin temperature of the right forearm and hand was measured before the period of ischemia, every 10 min during the period of ischemia, and after recovery, and was maintained between 32.4 and 34.5° C. Cooling was minimized by covering the hand and the forearm with a thick layer of cotton wool, and with a nearby infrared heater. After at least 2 mo of treatment with nasal continuous positive airway pressure (nCPAP), seven OSA patients were reevaluated in the manner just described. Supramaximal stimulation (mA), conduction velocity (m/s), potential amplitude (µV), and potential duration (ms) were studied. Latency was measured from the onset of the stimulus artifact to the onset of the action potential. Conduction velocity was obtained by dividing the distance between the stimulating and the recording electrodes by the latency. Potential amplitude was measured from peak to peak. Potential duration was measured from the onset of the potential to the positive peak (Figure 2).

View larger version (104K): [in this window] [in a new window]   Figure 1.   Ischemia test: Position of limb and arrangement of electrodes used for stimulating and recording evoked mixed nerve and sensory nerve action potentials.

View larger version (30K): [in this window] [in a new window]   Figure 2.   Representative tracings of mixed nerve conduction before, during, and 6 min after a 30-min period of limb ischemia in control subjects, OSA-NR and OSA-RICF patients. Note the disappearance of nerve conduction in control subjects and in OSA-NR patients.

Statistical Analysis

Results are expressed as mean ± SD, and were analyzed with nonparametric tests. The Mann-Whitney U-test was used to compare unpaired data (preischemic nerve conduction studies, population characteristics), and Wilcoxon's test was used to compare paired data (nerve conduction studies before and after ischemia, before and after treatment). Analysis of ischemia test data and comparisons of resistant with nonresistant subjects were done with the Kruskal-Wallis test, which was followed if necessary by a post hoc pairwise Mann-Whitney U test. Fisher's exact test was used to compare qualitative data. The level of significance was set at   0.05, and was adjusted with Bonferroni's procedure (difference statistically significant if p < 1  (1  )^1/k in the case of independent tests, or p < /k in the case of dependent tests, where k is the number of comparisons).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Characteristics of patients and control subjects are summarized in Table 1. In comparison with control subjects, the body mass index (BMI), Epworth Sleepiness Score, incidence of arterial hypertension, and nocturnal oxygen desaturation were higher in OSA patients, whereas their Sa_O2 during wakefulness was not statistically different.

Preischemic Nerve Conduction

Although the supramaximal stimulation required to obtain sensory and mixed nerve action potentials was higher in OSA patients than in control subjects, the amplitude was lower and the potential duration was higher in patients (Table 2). Sensory conduction velocity was significantly slower in OSA patients, whereas mixed nerve conduction velocity did not differ in the two groups.

Nerve Conduction during Ischemia

Mixed nerve action potential. Representative tracings of the mixed nerve action potential during ischemia are shown in Figure 2. In control subjects and in 10 OSA patients, the mixed nerve potential amplitude progressively decreased and finally disappeared at 24 min of ischemia, corresponding to classic ischemic conduction failure (ICF) (Figure 3, left panel). In the remaining seven OSA patients, mixed nerve conduction persisted during the 30-min period of ischemia, representing the resistance to ischemic conduction failure (RICF). Control subjects had the highest initial mixed nerve potential amplitude and greatest decrease in amplitude. The seven OSA-RICF patients had the lowest initial amplitude and the lowest decrease in amplitude of the mixed nerve action potential, with persisting nerve activity during ischemia. Although it was not statistically significantly different from that of the OSA-RICF patients except at 22 min of ischemia, the 10 OSA patients who were nonresistant to ICF (OSA-NR) had an intermediate response. Because the last minutes of the 30-min period of ischemia were often painful, we measured the time required for a 50% decrease in amplitude (T_50%) in order to study the possibility of an earlier discrimination between the different groups. However, T_50% was close for the three groups, at approximately 12 min for control subjects, 10 min for OSA-NR patients, and 14 min for OSA-RICF patients. Potential duration increased progressively during ischemia in the three groups. Patients had higher values than control subjects, whereas there was no significant difference between the two groups of OSA patients. Conduction velocity decreased during ischemia in the same manner for control subjects and OSA-NR patients, whereas OSA-RICF patients had the flattest slope. However, there was no significant difference among the three groups in conduction velocity.

View larger version (18K): [in this window] [in a new window]   Figure 3.   Mixed nerve action potential amplitude before, during, and after a 30-min period of ischemia before treatment (left panel ) for control subjects and OSA patients, and after at least 2 mo of treatment of OSA patients (right panel ). Results are expressed as the mean ± SEM. Analysis was done with the Kruskal-Wallis test, followed if necessary by a post hoc pairwise Mann-Whitney U test with Bonferroni's correction. Mixed nerve conduction disappeared at 24 min of ischemia in control subjects and in 10 OSA-NR patients, whereas seven OSA-RICF patients were resistant to ischemia. There was a significant difference (p < 0.0169) between control subjects and OSA-NR patients (*) and between control subjects and OSA-RICF patients (+). The difference between OSA-NR and OSA-RICF patients was significant only at 22 min. After treatment, OSA-RICF patients became nonresistant, whereas preischemic nerve conduction parameters of the two subgroups of patients were not significantly different from those measured before treatment.

Sensory nerve action potential. The amplitude of the sensory nerve action potential progressively decreased during ischemia in the three study subgroups, but was reduced by only 33.5% in OSA-RICF patients. As with the mixed nerve potential, controls had the highest initial amplitude and the highest slope, whereas OSA-RICF patients had the lowest initial amplitude and the lowest slope. OSA-NR patients, who did not differ significantly from control subjects and OSA-RICF patients in their sensory nerve action potential amplitude, had an intermediate response. T_50% was 22 min for control subjects and OSA-NR patients, whereas the decrease in amplitude for OSA-RICF patients never reached 50%. Potential duration increased progressively in the three groups. OSA patients had higher values than control subjects, whereas there was no difference between the two subgroups of OSA patients. Control subjects had a higher conduction velocity than did the OSA patients. Although it was not significantly different from that of OSA-NR patients, OSA-RICF patients had the lowest values for conduction velocity. Conduction velocity decreased in the same manner for control subjects and OSA-NR patients, whereas OSA-RICF patients had the flattest slope.

Nerve Conduction during Recovery

The mixed nerve action potential amplitude was not significantly different from its preischemic values at 2 min after ischemia for OSA-RICF patients (4.6 ± 1.5 µV versus 8 ± 1.7 µV) (postischemic value versus preischemic value), whereas it remained lower at 6 min for control subjects (18.3 ± 9.8 µV versus 28 ± 14.4 µV) and OSA-NR patients (7.8 ± 3.2 µV versus 14 ± 6.4 µV). However, the postischemic action potential amplitude of control subjects was the highest, whereas the amplitudes in the two OSA patient subgroups were not significantly different (Figure 3, left panel). The sensory nerve action potential amplitude was not significantly different from its preischemic values at 2 min for the two subgroups of patients (OSA-NR: 19.5 ± 16.5 µV versus 35.1 ± 20.8 µV; OSA-RICF: 13.2 ± 7.4 µV versus 16.8 ± 6.7 µV), or from its preischemic value at 6 min for control subjects (36.7 ± 19.4 µV versus 42.9 ± 16.3 µV). Only for control subjects and OSA-RICF patients were the mixed nerve action potentials significantly different at 6 min. Compared with their preischemic values, the durations of sensory and mixed nerve potentials remained significantly greater at 6 min for control subjects (sensory: 1.78 ± 0.21 ms versus 1.44 ± 0.14 ms; mixed: 1.7 ± 0.39 ms versus 1.34 ± 0.15 ms), whereas the potential durations were not significantly different at 2 min for OSA-RICF patients (sensory: 2.38 ± 0.76 ms versus 1.89 ± 0.34 ms; mixed: 2.14 ± 0.35 ms versus 1.54 ± 0.18 ms), and at 6 min for OSA-NR patients (sensory: 2.1 ± 0.39 ms versus 1.79 ± 0.33 ms; mixed: 1.99 ± 0.36 ms versus 1.51 ± 0.2 ms). The durations of sensory and mixed nerve potentials in the three groups of subjects were not significantly different at 6 min. Conduction velocity of sensory and mixed nerve potentials was not statistically different from preischemic values at 2 min for OSA-RICF patients (sensory: 38.1 ± 7.2 m/s versus 42.1 ± 5 m/s; mixed: 55.4 ± 5.8 m/s versus 58.9 ± 3.3 m/s), whereas it remained significantly lower at 6 min for control subjects (sensory: 47.1 ± 2.8 m/s versus 51.4 ± 2.8 m/s; mixed: 55.5 ± 1.3 m/s versus 61.7 ± 2.6 m/s) and OSA-NR patients (sensory: 39.5 ± 4.6 m/s versus 45.6 ± 5.6 m/s; mixed: 55.8 ± 5.2 m/s versus 63 ± 3.4 m/s). However, the sensory conduction velocity of control subjects was the greatest, whereas for OSA-RICF and OSA-NR patients there was no significant difference. The mixed nerve conduction velocity in the three groups was not statistically significantly different.

Differences between Control Subjects, OSA-NR, and OSA-RICF Patients

Control subjects had a smaller arm circumference than did OSA-RICF patients, whereas there was no difference between control subjects and OSA-NR patients or between the two groups of patients (Table 3). On the other hand, OSA-RICF patients had a higher BMI than either OSA-NR patients or control subjects. Hypertensive subjects were present only in the two patient groups. Although OSA-RICF patients had the highest values, blood pressure was not significantly different in the three groups. Mean oxygen saturation during wakefulness was also not significantly different among the three groups. However, mean nocturnal oxygen saturation was lowest in the OSA-RICF group. Furthermore, the cumulative duration of oxygen desaturation was greater in OSA-RICF than in OSA-NR patients. There was a trend toward greater desaturation in OSA-RICF patients in terms of the Sa_O2 from < 90% to < 75%. This became significantly different when comparing the duration of an Sa_O2 < 70% and lower. The AHI and microarousal index were not significantly different in the OSA-RICF and OSA-NR patients.

Nerve Conduction After Treatment

Four OSA-RICF patients and three OSA-NR patients were reevaluated after at least 2 mo of treatment with nCPAP (136 ± 62 d for the OSA-NR patients and 143 ± 89 d for the OSA-RICF patients) (Figure 3, right panel). Nerve conduction parameters before ischemia (amplitude, duration, and conduction velocity of sensory and mixed nerve action potentials) were not significantly different from their pretreatment values for the two groups of patients, whereas resistance to ischemia disappeared in the four OSA-RICF patients.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study described here is the first to have demonstrated peripheral nerve dysfunction in OSA patients. This was characterized by abnormal baseline electrophysiologic activity, which for some patients persisted during ischemia. Preischemic distal sensory and mixed nerve action potential amplitudes were lower in OSA patients than in control subjects despite a higher supramaximal stimulation, suggesting the presence of axonal lesions. In addition, there was an increase in both sensory and mixed nerve potential duration, expressing an increased temporal dispersion of conduction; and a reduced sensory conduction velocity, suggesting lesions to myelin. Thus, the preischemic data for OSA patients indicate peripheral neuropathy that was principally axonal. This nerve dysfunction was more pronounced and associated with RICF in those patients with the most severe nocturnal hypoxemia. Taking into account the clinical characteristics of OSA patients, electrophysiologic abnormalities might be related to obesity or hypertension. However, the disappearance of RICF upon treatment with nCPAP, without a change in BMI or daytime blood pressure, makes a significant contribution of obesity or hypertension to RICF unlikely. With regard to axonal lesions, obesity and hypertension are not known to induce peripheral polyneuropathy. However, hyperinsulinemia combined with insulin resistance states such as obesity may contribute to nerve dysfunction. Indeed, insulin administration could impair tissue oxygenation through a deleterious vasoactive effect (22). Although we excluded diabetic subjects from our study, insulinemia and insulin resistance were not evaluated in our patients. Similarly, because hypertension is known to induce cerebral lesions through atherosclerosis and small dissecting aneurysms, its contribution to nerve lesions through vascular mechanisms cannot be excluded. Because OSA-RICF patients had the largest arm circumference, the decrease in mixed nerve action potential amplitude in these patients may be in part attributed to their subcutaneous fat thickness, which could have interfered with the recording of the action potential amplitude at the elbow. However, the decrease in amplitude was also noted with the sensory nerve action potential, which was recorded at the digit, and was therefore unrelated to any artifact produced by subcutaneous fat. Because subcutaneous fat thickness does not interfere with the recording of potential, duration or conduction velocity, the abnormalities noted in these two parameters express pathologic nerve conduction without ambiguity.

Axonal Lesions

COPD is well known to induce peripheral neuropathy, which is related to the severity and duration of hypoxemia (17). Both hypoxic and diabetic neuropathy are associated with nerve capillary endothelial-cell hyperplasia and hypertrophy, predisposing to luminal occlusion, combined with thickening of the nerve perineurium, which may impede the transport of nutrients and oxygen (18). In diabetic neuropathy, nerve lesions may be caused by a hypoxic process (18) and oxidative stress (23) through hyperglycemic pseudohypoxia (24). Likewise, in hypoxia-related neuropathy, the role of oxidative stress may be predominant. It has been shown that isolated nocturnal hypoxia induces degradation of adenosine monophosphate (AMP) and production of hypoxanthine, xanthine, and uric acid (10). These metabolic products of adenine nucleotides lead to the production of free oxygen radicals during reoxygenation. By analogy with what occurs during ischemia and reperfusion, the following sequence may occur during severe transient hypoxemia (25): stimulation of the glycolytic pathway and development of tissue acidosis with accumulation of reduced nicotinamide adenine dinucleotide (NADH), citrate, and lactate; intracellular sodium accumulation with continuing hypoxia and therefore energy deficit; intracellular calcium accumulation during reoxygenation through exchange of calcium with sodium; and calcium-dependent activation of phospholipases, proteases, and endonucleases, leading to damage to the sarcolemma and cytoskeleton, thus contributing to free-radical generation, cytoplasmic bleb formation, DNA strand breaks, and increased expression of immediate/early genes, which may be important in initiating cell death or apoptosis. Most if not all of these mechanisms may be applicable to peripheral nerve dysfunction and lesions, resulting from impaired axonal transport (an energy-requiring process) to axonal degeneration. Furthermore, the endothelial lesions related to hypoxemia may contribute, along with functional and structural alterations of the microvasculature in OSA patients (26), to nerve lesions by reducing nutrient and oxygen transport.

Excitotoxicity mediated by excitatory amino acids such as glutamate has been reported in hypoxic CNS injury (27). This excitotoxicity seems to be closely related to oxidative stress (28), impaired energy metabolism, and neurodegeneration (29), but its role in peripheral nerve pathology remains to be established.

High plasma levels of insulin are commonly found in OSA patients (30). Alterations in intracellular ion concentrations such as increased free cytosolic calcium, depletion of free magnesium, and low pH, have been observed in various states of insulin resistance, such as non-insulin-dependent diabetes mellitus and obesity (31). Because these intracellular abnormalities are also found in hypoxemia, we postulate that they may be a common pathway for the pathologies associated with OSA, such as hypertension, insulin resistance, and peripheral nerve dysfunction.

RICF

The meaning and the physiopathology of RICF remain unclear. The occurrence of RICF in various metabolic conditions that are known to induce peripheral neuropathy, such as diabetes (12), chronic hypoxia (12, 13), and chronic renal failure (32), and also in chronic hepatic failure (33), hypercalcemia (34), and with increasing age (35), suggests that RICF is a nonspecific phenomenon that may be induced by different mechanisms. In experimental diabetes, the increase in endoneural substrates (glucose, fructose, sorbitol, and glycogen), and anaerobic glycolysis secondary to endoneural hypoxia and reduction of energy requirements, may largely explain RICF (36). RICF related to experimental diabetes is reduced within 1 h after insulin administration (37), and also by 4 wk of oxygen supplementation (15). RICF can also be produced in rats by hyperglycemia alone, without the presence of diabetic neuropathy (38). The difference in effect of insulin and oxygen treatment may be explained by the rapid correction of hyperglycemia with insulin and by the correction with oxygen treatment of the endoneural consequences of hypoxia related to diabetes. Conversely, hyperglycemia induces early RICF (within 1 h) (37), whereas 4 wk exposure to normobaric 10% hypoxia are needed to induce RICF (39).

Identifying RICF in OSA may help in understanding the underlying physiopathology of RICF. It should be noted that blood oxygen content is normalized in OSA-RICF patients at the end of the night. Thus, there may be either structural or metabolic changes persisting during the day, at least at the peripheral nerve level. With chronic and continuous hypoxia there are two possible ways to produce RICF. One is to reduce energy requirements (39) and to optimize the use of substrates. Indeed, during cellular hypoxia, as the consumption of lactate ceases, glucose rises to 70 to 90% of the total substrate consumed. By increasing the rate of glucose metabolism, the hypoxic cell becomes a more efficient user of O₂ because the metabolism of glucose yields more moles of adenosine triphosphate (ATP) per mole of O₂ consumed than does the metabolism of either fat or protein. The second way in which RICF may be produced is to increase the anaerobic production of ATP through anaerobic glycolysis, creatine kinase and adenylate kinase reactions, and catabolism of adenine nucleotides (25). Other general adaptations to hypoxic conditions, such as an increase in hemoglobin, improving the oxygen carrying capacity of the blood, are unlikely to be important; there was no detectable polycythemia in our patients. That we observed RICF during the daytime, when blood oxygen content had been normalized for several hours, indicates that several or all of these biochemical changes may rely on durable adaptations of the nerve, at a distance from the acute exposure to intermittent nocturnal hypoxia. The possibility of long-lasting metabolic changes as leading to RICF is strengthened by the progressive appearance of RICF only after 4 wk of normobaric hypoxia (39).

Axonopathy and RICF

In our study, we observed either axonopathy or axonopathy plus RICF, but not isolated RICF. Our results therefore suggest that chronic exposure to severe intermittent hypoxia resulting in RICF is always associated with axonopathy. In the RICF patient subgroup, the baseline amplitude of the action potential was lower than in the other patients. This suggests that RICF coexists with axonal lesions of greater severity than those of patients without RICF. RICF and axonal lesions were related to the severity of nocturnal hypoxemia. We can advance two hypotheses for the occurrence of RICF with axonopathy, which could be interlinked. First, RICF may depend on a critical threshold of total nocturnal oxygen available to tissues, which arises from severe nocturnal oxygen desaturation and microvascular disorders. Second, the coexistence of RICF plus severe axonal lesions may represent a progression of OSA-related disease, although its duration is impossible to determine. Indeed, in experimental diabetes, Nukada (40) observed RICF without morphologic vulnerability at 2 wk after the induction of diabetes. After 16 wk, RICF and morphologic changes in nerves coexisted. At this time, the administration of insulin reduced RICF without affecting morphologic changes. In our study, the disappearance of RICF with persisting axonal neuropathy during treatment with nCPAP closely resembles the experimental observation of Nukada (40), and is in favor of the first hypothesis.

Conclusions

OSA patients have two types of nerve dysfunction: (1) preischemic abnormalities of nerve conduction, which mainly reflect axonal degeneration; and (2) RICF, which appears to be a sensitive but nonspecific tissue marker of a severe metabolic disorder (severe nocturnal hypoxemia in our study), and which may result from adaptative mechanisms. The common pathway between glucose intolerance and the metabolic disorders related to hypoxemia may not be restricted to peripheral nerves. Understanding of the mechanisms responsible for RICF in OSA may require the use of experimental models of OSA, allowing biochemical and pharmacologic studies of this disorder.