INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The patency of the upper airway depends on the balance between the negative intrapharyngeal pressure developed during inspiration and the counteraction of the dilating muscles (1). In patients having obstructive sleep apnea (OSA), the upper airway collapses during inspiration and sleep, most commonly at the pharyngeal level (2). The subocclusive stage of habitual snoring usually precedes the development of OSA, but the pathogenesis of progression has not been fully clarified. Lugaresi and coworkers proposed that there is a progressive heavy snorers disease (3). Support for this theory is given in a study by Svanborg and Larsson (4), who found worsened results in repeated respiratory recordings of untreated habitual snorers.

Several studies in humans have demonstrated that the negative pharyngeal pressure caused by inspiration activates both velo- and oropharyngeal muscles (5), with a decreased muscle activity during sleep (6). Our hypothesis is that a progressive lesion in the afferent and/or efferent nerve pathways in this reflexogenic mechanism, caused by the snoring trauma, is a contributory factor to the collapsibility and obstruction of the upper airway seen in patients with OSA. Evidence for the existence of a local pharyngeal sensory neuropathy has been reported since impaired temperature sensitivity is present in OSA patients (7). Further, electron microscopy of the uvula in patents with severe OSA has shown focal degeneration of myelinated nerve fibers (8).

Our research group has previously shown that signs of a peripheral nervous lesion were present in histological preparations of the palatopharyngeus muscle from OSA patients, whereas no such signs were found in biopsies from nonsnoring control subjects (9). However, that study was purely descriptive concerning the morphological abnormalities. The aim of the present study was to evaluate the hypothesis of a progressive neurogenic lesion in an upper airway muscle in patients with different degrees of upper airway obstruction. Therefore, newly cut sections were investigated from previous specimens (seven nonsnoring control subjects and seven OSA patients), and an addition was made of specimens from new subjects (three control subjects, three OSA patients, and 11 snoring patients without laboratory criteria of OSA). The morphological abnormalities were this time blindly semiquantified for statistical comparisons between nonsnorers and patients with habitual snoring but different degrees of upper airway obstruction.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Thirty-one men undergoing pharyngeal surgery were included in the present study. In 21 patients, uvulopalatopharyngoplasty (including tonsillectomy) was performed as treatment of habitual snoring (with or without OSA) and various degrees of excessive daytime sleepiness. The patients had a median age of 47 yr (range 33-62) and a median BMI of 28.4 (21.7-36.3) kg/m². Patients with gross abnormalities of the upper airway (i.e., severe tonsillar hyperplasia, micrognathia) or with previous treatment for their snoring/OSAe.g., CPAP, dental devices were excluded.

Ten nonsnoring men served as control subjects. They all underwent tonsillectomy because of chronic tonsillitis, except for two men who underwent the operation as part of a search for a primary tumor, which was subsequently found elsewhere. The control subjects had median age of 36.5 yr (17-71) and a median BMI of 24.4 (range 18.7- 28.1) kg /m².

None of the subjects had clinical signs or symptoms of a polyneuropathy or any neuromuscular disorder. Apart from three patients using antihypertensive drugs, none was on medication.

All subjects had preoperatively given their informed consent to participation in this study, which was approved by the local ethics committee.

Respiratory Sleep Recordings

All patients had preoperatively undergone a whole-night sleep recording, including monitoring of breathing and body movements by means of a sleep apnea mattress (PVDF) and pulse oximetry (BIOX 3740, ear probe; OMEDA Inc., Louisville, CO). This method has previously been validated against full polysomnography (10). From these recordings, the oxygen desaturation index (ODI, average number of  4% desaturations per hour of sleep), the lowest (nadir) Sa_O2 and the percentage periodic breathing of obstructive type of total sleep time were calculated. Periodic obstructive breathing is here defined as at least three consecutive cycles of waxing-waning respiratory movements, where the highest amplitude in each cycle must exceed the lowest by at least 50%. According to the previous validation study (10) the criteria for OSA are ODI > 6 and > 45% periodic obstructive breathing. Recordings with ODI < 2 and < 18% periodic obstructive breathing are considered to be normal. Recordings with ODI  2 and  18% obstructive breathing without fulfilling the criteria for OSA are in a border-line state. However, a polysomnography could eventually have revealed a slight to moderate OSA in border-liners (10).

The 10 OSA patients (nos. 12-21) had a median ODI of 32 (range 7-80), a median percentage obstructive breathing of 73% (45-99), and a median nadir of 74% (47-85). Six patients (nos. 1-6) with normal recordings had a median ODI of 0 (0-1), a median percentage obstructive breathing of 4.5% (0.9), and a median nadir of 92% (89-94). The five patients (nos. 7-11) with border-line recordings had a median ODI of 3 (0-3), a median percentage obstructive breathing of 19% (16-74), and a median nadir of 88% (88-91). Four patients (nos. 6, 12, 17, and 18) also underwent full-night conventional polysomnography (PSG) which confirmed their diagnosis.

The nonsnoring control subjects had not undergone any sleep recordings but were thoroughly interviewed by one of the authors (HL or DF), for any history of snoring, which they and/or their spouses denied. For individual data of respiratory recordings, see Table 1.

Biopsy Procedures

The biopsies, with an approximate size of 10 × 5 × 5 mm, were obtained from the same cranial portion of the palatopharyngeal muscle, exposed during surgery. All surgery was performed under general anesthesia without use of local anesthetics, except in one patient (no. 10), in whom lidocaine 5 mg/ml with epinephrine 5 µg/ml was injected locally.

In addition, percutaneous control biopsies from anterior tibialis muscle were taken to investigate whether the abnormalities found in the palatopharyngeal muscle were signs of a general or a local lesion. These biopsies were obtained under local anesthesia (lidocain 5 mg/ ml with epinephrine 5 µg/ml) from Patients 9, 10, 13, and 14 (see Table 1).

The muscle specimens were quickly frozen in freon-13, cooled with liquid nitrogen (190° C) and stored at 75° C until further processed. Sections of 10 µm were cut in a cryostat operating at 25° C. Other sections of specimens of the palatopharyngeus muscle from seven patients (nos. 12, 15, and 17-21, see Table 1) and seven controls (nos. 3, 5-10) had previously been investigated (9). New-cut sections of these 14 frozen specimens underwent similar preparations simultaneously with the sections from the newly recruited 17 subjects. Cryosections were stained with hematoxylin-eosin (11), modified trichrome (12), NADH-tetrazolium reductase (NADH-TR) (13), a marker for oxidative enzyme activity, and myofibrillar adenosine triphosphatase (ATPase) (14). The classification of muscle fiber types was based on their ATPase-staining characteristics (15).

Morphological Analysis

The evaluation of the morphological abnormalities was made by two myologists (TA, KB) in concert, which is a routine procedure. These myologists were not the same as in the previous study (9). The blinded cryosections were analyzed concerning identification of unspecific abnormalities (see RESULTS) as well as specific signs of neurogenic lesions, i.e., type-grouping, grouped atrophy and fascicular atrophy, changes which indicate different degrees of severity of denervation and reinnervation (16, 17). According to a summation of all such abnormalities, each specimen was classified into one of four categories: 0 = normal morphology, 1 = slight, 2 = moderate, and 3 = severe pathology.

In addition, to verify this classification of abnormalities made in concert, the interindividual variation between the myologists was measured. Thirty-one blinded random sections (stained for ATPase) were reevaluated separately by the two investigators.

Morphometrical Analysis

For morphometrical analysis of palatopharyngeal muscle, the fibers were measured directly from the microscope via a CCD camera (Hamamatsu C3077; Hamamatsu Photonics KK, Joko-Cho, Hamamatsu City, Japan) connected to an image-analysis processor (Vidas; Kontrol Bildanalyse, GmbH, Munich, Germany).

In analyzing the proportion of fiber types, the cryosections stained with ATPase at pH 4.6 and 4.3 were photographed (×100) and printed out on paper. The numbers of fiber type I, IIA, IIB and IIC, respectively, were counted and expressed as percentage of the total number of fibers.

The cross-sectional areas and lesser diameters of the fibers were measured on 100 type I and type IIA fibers and on 50 type IIB fibers from each biopsy specimen. If the total numbers of the respective type were less than 100 or 50, then all the fibers of that type were measured. If the total numbers of fibers were very low (less than 25), they were not included in the statistical evaluation. The individual mean cross-sectional area of the fibers of palatopharyngeal muscle was calculated in order to investigate the occurrence of atrophic and hypertrophic fibers, which is not revealed by the mean values. A healthy muscle has a relatively uniform fiber-size distribution, but in the palatopharyngeus muscle of OSA patients, the distribution has been described as bi- or polymodali.e., hypertrophic and/or atrophic fibers (9). The individual histograms were blinded and classified by three of us (TA, KB, HL) into two categories: 1 = normal-to-slight polymodality and 2 = moderate-to-severe polymodality. Those histograms in which the evaluation differed between the investigators were classified according to the evaluation of at least two of three investigators, i.e., the principle of majority.

Immunohistochemical Procedures

Immunolabeling with antibodies against the cytoskeletal proteins dystrophin, spectrin, desmin, and vimentin were performed to investigate if the snoring trauma affected the cytoskeletal structure. Antibodies against Leu-19, a myoblast/satellite cell antigen, used as a marker for muscle fiber regeneration (18) were added. The cross-sections of palatopharyngeus muscle were labeled with commercial mouse monoclonal antibodies directed against the cytoskeletal proteins dystrophin (NCL-DYS 2 and NCL-DYS 3; Novocastra Laboratories, Ltd., Newcastle Upon Tyne, UK), desmin (Clone 33; Monosan, Uden, Netherlands), beta-spectrin (NCL-SPEC 1 and NCL-SPEC 2; Novocastra Laboratories, Ltd., Newcastle Upon Tyne, UK) and vimentin (Monosan), and also Leu-19 (CD-56; Dakopatts a/s, Glostrup, Denmark). The second antibody was a biotinylated sheep anti-mouse antibody, which was detected using an avidin-biotin complex kit (Vector Laboratories Inc., Burlingame, CA). The avidin-biotin complex was visualized by means of 3.9-aminoethylcarbazole (Sigma Chemical Co., St. Louis, MO).

Statistical Analysis

Nonparametrical methods were used. Comparisons between the control group and the entire group of patients were made with the Mann-Whitney U test. Spearman rank correlation test was used for correlation tests between variables.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Morphology

The palatopharyngeal muscle of the control subjects exhibited slightly greater variations in fiber size, with occurrence of rounded atrophic fibers (Figure 1A), a slightly increased incidence of centrally located nuclei and split fibers, as compared with what is normally seen in limb muscles. The fibers often displayed irregular staining with hematoxylin-eosin (Figure 1B).

View larger version (133K): [in this window] [in a new window]   Figure 1.   Cross-sections of the palatopharyngeal muscle from a control subject. The sections were stained with hematoxylin-eosin (A) and, for myofibrillar ATPase, after acid preincubation at pH 4.6 (B). In (A), note the variation in fiber size and the occurrence of small rounded fibers and of fibers with irregular staining characteristics. In (B), normal types of fibers with a predominance of type II fibers are seen. Type I fibers stain black, type IIA fibers light, and type IIB fibers stain grey. Bar = 50 µm.

The specimens of the entire group of patients exhibited extensive morphological abnormalities, including signs specific for neurogenic lesionse.g., type grouping, fascicular atropy and/or grouped atrophy. There was an increased occurrence of nonspecific changes, such as rounded or angulated, atrophic fibers (Figure 2A), hypertrophic and/or split fibers, in addition to fibers with centrally located nuclei and/or sarcoplasmic masses. Further, when stained for oxidative enzymes (NADH-TR), a large number of muscle fibers were found to have an irregular internal structure, including moth-eaten (Figure 2B), whorled and/or target fibers. The morphological abnormalities were not confined to a specific fiber type.

View larger version (119K): [in this window] [in a new window]   View larger version (150K): [in this window] [in a new window]   Figure 2.   Cross-sections of the palatopharyngeal muscle from two OSA patients. The sections were stained with myofibrillar ATPase at pH 9.4 (A) and NADH-TR (B). In (A), type II fibers stain black and type I fibers gray. Note the predominance of type II fibers. Arrows point to a fascicle with atrophic fibers, indicating a neurogenic process. In (B), note the irregular staining pattern and the abundance of moth-eaten fibers. Bar = 100 µm in (A) and 50 µm in (B).

The results of the classification of abnormalities, with comments of the dominating specific signs, are shown in Table 1. The degree of abnormality was significantly increased in the entire group of patients compared with control subjects, p < 0.001 (Mann-Whitney U test). The specimens from a control subject exhibited the slightest changes, whereas a moderate-to-severe pathology was observed in 12 of 21 (57%) patients, i.e., seven of 10 OSA patients, four of five borderliner patients, and one of six patients with normal recordings. None of the patient specimens was normal.

There was no significant correlation between the degree of abnormality and ODI, nor with lowest (nadir) Sa_O2. However, the degree of abnormality and percentage periodic obstructive breathing in patients correlated significantly, p = 0.04 (Spearman rank correlation test), (Figure 3). The patients were therefore divided into two groups according to the laboratory criteria for normality versus pathological upper airway obstruction, in terms of periodic obstructive breathing. One group consisted of nine patients (nos. 1-9) with < 20% obstructive breathing and one group of 12 patients (nos. 10-21) with  45% obstructive breathing (Table 1). The degree of abnormality correlated significantly with the three subject groups (i.e., nonsnoring controls, snoring patients with < 20% periodic breathing, and snoring patients with  45% periodic breathing), p < 0.001 (Spearman rank correlation test), as shown in Figure 4.

View larger version (23K): [in this window] [in a new window]   Figure 3.   The significant correlation between degree of abnormality (slight, moderate and severe) and percentage obstructive respiration in patients with habitual snoring. No specimen was normal among the patients. Box plot: the boxes indicate the median values with 25 and 75 percentiles, the bars indicate 10 and 90 percentiles, in each group. Individual values outside these percentiles are indicated (round symbols).

View larger version (28K): [in this window] [in a new window]   Figure 4.   The significant correlation between degree of morphological abnormality (0 = normal, 1 = slight, 2 = moderate, 3 = severe pathology) and the three subject groups (nonsnoring control subjects, patients with < 20% obstructive breathing and patients with 45% obstructive breathing). Box plot: the boxes indicate the median values with 25 and 75 percentiles, the bars indicate 10 and 90 percentiles, in each group. An individual value outside these percentiles is indicated (round symbol ).

Morphological signs specific for neurogenic lesions (type grouping, fascicular atrophy and/or grouped atrophy) were found in the specimens of 15 of 21 patients (71%) (see Table 1 and Figure 2A) compared with only slight type-grouping or fascicular atrophy in two of 10 controls. In addition, moth-eaten and target fibers, often seen in connection with neurogenic muscle disorders, were seen in a majority of the patients, but in only one of the controls.

In order to establish the interindividual variation of the scoring of pathology, the results of the two myologists were compared. They varied at the most with one category (e.g., slight versus moderate) in five of 31 sections. The figures given in Table 1 are results of their scoring in concert.

The biopsy specimens of the anterior tibial muscle were normal in all patients, except in one (no. 13), who demonstrated slight unspecific changes, but no signs of neurogenic lesions.

Morphometry

The number of counted fibers in each specimen from palatopharyngeus ranged from 180 to 900, with an average of 600 fibers. The fiber type distribution did not differ between controls and patients. The majority of fibers were of type IIA (median 67%, range 1-92), followed by type I (median 26%, 4-98), type IIB (median 2%, 0-20), an type IIC (median 1%, 0-7). In contrast, two control subjects (nos. 8 and 9) and three patients (no. 4, 11, 16) had a type I-fiber predominance.

The analysis of the individual mean lesser diameters and mean cross-sectional areas of the fibers did not reveal any significant differences between the groups. On the other hand, the individual fiber-size histograms (= spectrum) did. All specimens were evaluated concerning the histograms, except in one patient (no. 12), whose spectra had too few fibers of both type I and type IIA-fibers to allow evaluation. Six spectra of 30 specimens also contained too few fibers (less than 25) of one of these types, and generally the type IIB and IIC fibers. In total, 54 spectra, 28 of type I- and 26 of type IIA fibers were evaluated.

All spectra from the 10 control specimens were normal or slightly biomodal, except in one (no. 2), in which the spectrum of type I fibers was moderately bimodal. Fifteen of the 20 (75%) patient specimens contained moderate-to-severe pathological spectra of either type I, type IIA fibers or both. The degree of pathological spectra was significantly increased in patients compared with control subjects concerning type I as well as type IIA, p < 0.01, 0.002 (Mann-Whitney U test), respectively. This degree also correlated significantly to subject group concerning both fiber types, p < 0.05 and < 0.005, respectively, as shown in Figure 5. There were only minor differences between the patient groups, and the degree of pathological spectra did not correlate to ODI or percentage obstructive breathing.

View larger version (32K): [in this window] [in a new window]   Figure 5.   Diagram of the number of patients with normal to slight, and moderate to severe pathological fiber-size spectra of type I (n = 28), and type IIA fibers (n = 26) in the three subject groups, i.e., nonsnoring controls, snoring patients with < 20% periodic obstructive breathing, and snoring patients with  45% obstructive breathing (10 OSA patients). Note the significant correlation between subject group and degree of pathology.

In 12 of 54 (22%) spectra, four of type I and eight of type IIA, the categorization differed between the three investigators. Discarding those spectra due to the discordance would not change the significant differences between control and patient group for type I or type IIA, p < 0.01 and < 0.05, respectively (MWU-test).

Immunohistochemistry

A normal subsarcolemmal staining pattern for dystrophin and spectrin was noted in the majority of fibers, irrespective of group. A small number of fibers in patient specimens exhibited invaginations and infolded loops of the sarcolemma. The majority of fibers showed a normal staining for desmin, with typical cross-striations in the controls and in both patient groups. However, a small number of atrophic and normal-sized fibers showed an abnormal staining pattern for desmin, with only a subsarcolemmal rim of staining, while the sarcoplasm was very faintly stained or was unstained (Figure 6A). The number of such rim-fibers was slightly larger among the patients than in controls. No muscle fibers in any specimens were positively stained with vimentin.

View larger version (131K): [in this window] [in a new window]   View larger version (111K): [in this window] [in a new window]   Figure 6.   Cross-sections of the palatopharyngeal muscle from two OSA patients. The sections were stained for desmin (A) and Leu-19 (B). In (A), there are a few fibers with a subsarcolemmal rim of staining whereas the sarcoplasm remains unstained. Some of these fibers are indicated with asterisks. In (B), several fibers stain positively with Leu-19 (some of which are indicated with arrowheads), suggesting regeneration of these fibers. Bar = 100 µm in (A) and 25 µm in (B).

In control subjects and patient specimens, a small number of atrophic, but also normal sized fibers, stained positively for Leu-19 (Figure 6B). However, the number of positively stained fibers had increased, as judged qualitatively, in the patient specimens, as compared with the control subjects.

Age did not significantly correlate to degree of abnormality, nor to degree of pathological fiber spectra. Body mass index (BMI) correlated significantly to degree of abnormality, p < 0.01 (Spearman rank correlation test), but not to pathology of fiber spectra.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main findings in this study were (1) the significant increase of morphological abnormalities, including signs characteristic of neurogenic lesions, in the palatopharyngeal muscle of the entire group of snoring patients, compared with the nonsnoring controls; and (2) the percentage periodic obstructive breathing correlated significantly to the quantified morphological abnormalities. This indicates that the periodic obstructive breathing time of total sleeping time may be a quantitative measure of the magnitude of the snoring trauma to the pharyngeal tissues.

Interestingly, two of the patients with  45% obstructive breathing but low ODIs had a severe degree of abnormality. In the entire group of snoring patients, there was no correlation between oximetry findings and degree of abnormality It may be that ODI is a more qualitative index of the snoring trauma, or that it also reflects other factors than obstruction. Thus, the results indicate that patients with habitual snoring and high amounts of periodic obstructive breathing have an increased risk to achieve local muscular abnormalities despite fairly normal oximetry findings.

It is not likely that age is the cause of the abnormalities seen in the present study, since age did not significantly correlate to the abnormalities and one control was 71 yr. Obesity is neither a likely cause, since a majority of patients with abnormalities were not obese (BMI < 30). Instead our results indicate that, as no general neuropathy could be verified in the limb muscle, a local trauma could be a cause. The palatopharyngeal muscle is of anatomical importance in the pharynx as it forms the internal longitudinal muscular layer around the wall of the pharynx. The muscle is located at the major site of obstruction in OSA patients (2) and is therefore exposed to the vibration and stretch trauma induced by heavy snoring and obstructive breathing. Snoring represents a low-frequency vibration trauma (19), which may cause peripheral nerve injury (20).

An unexpected but interesting finding in the present study was that similar specific neurogenic signs in addition to the polymodal fiber-size specific neurogenic signs in addition to the polymodal fiber-size spectra (regarded as an expression of a neurogenic lesion [11]) were demonstrated in the snoring patients regardless of their degree of upper airway obstruction. Further, one snoring patient with a normal respiratory recording demonstrated a severe degree of muscular abnormality. Our results suggest that, in vulnerable patients, the vibration trauma of habitual snoring itself could initiate a local neurogenic lesion, before the additional trauma of stretch caused by periodic obstructive breathing.

A majority of the muscle fibers stained normally for the cytoskeletal proteins dystrophin, spectrin, vimentin, and desmin irrespective of group in the present study. This indicates that the cytoskeleton of pharyngeal muscle fibers is well preserved in snoring patients with or without OSA. However, a small number of fibers in snoring patients showed increased wrinkling of the sarcolemma, which has previously been reported in association with advanced age and denervation (21). In all specimens, a small number of fibers exhibited an abnormal staining pattern for desmin with staining only of the subsarcolemmal regions, leaving the center of the fiber weakly stained or unstained. The nature and significance of these rim-fibers is not yet known. Leu-19 is a marker for satellite cell activation and muscle fiber regeneration (18). The slightly increased occurrence of Leu-19 positive fibers in patient specimens is interesting since it has also been reported in other patients with various lower motorneuron disorders, including prior poliomyelitis (22) and compression of spinal roots (L. Edström, personal communication).

Concerning the fiber type proportion, our results are in accordance with those of Séries and coworkers (23), who found a predominance of type IIA of different degrees in uvula and genioglossus muscles in sleep apneics and snorers. Nonsnoring controls were not investigated. Smirne and coworkers (24) also found a type IIA predominance in the medium pharyngeal constrictor muscle in four patients with anamnestic snoring, but a predominance of slow-twitch type I fibers in nine nonsnoring controls. This latter finding is in contrast to ours of a fast-twitch type IIA predominance in nonsnoring subjects. There may be different explanations for this discrepancy e.g., the small number of control subjects and that two different muscles were investigated. The proportion of fiber types is primarily genetically determined (25). However, it is described in a rat study that a transformation between slow- and fast-twitch fiber types can occur in response to changes in the pattern and/or level of muscle activation (26). Our findings, based on enzyme histochemistry, do not indicate any such transformation in the palatopharyngeal muscle since the proportions of different fiber types were unchanged.

In the above quoted study by Smirne and coworkers, the authors stated that no neurogenic changes were found in the pharyngeal constrictor muscle (24). However, no histopathological results were described in the four snorers, except that the type IIA fibers were hypertrophied. Such a finding does not, however, exclude a neurogenic lesion. Fiber atrophy is the classic finding in neurogenic lesions, but neurogenic muscle hypertrophy has been demonstrated in patients with peripheral nerve lesion, according to a recently published review (27). In the present study, the mean values of the fiber size did not reveal any significant differences between controls and patients. However, analyses of the fiber-size distribution revealed that a majority of the patient specimens contained hypertrophied and/or atrophied fibers.

Muscle hypertrophy in neurogenic lesions has been suggested to be due to two factors (1) chronic stretch of innervated and denervated muscle fibers and (2) overuse of partially denervated muscles (27). During snoring and apneas the muscle fibers are subjected to stretch as the pharynx undergoes a caudal traction (2). Further, there are indications of an overuse of pharyngeal muscles in an effort to maintain upper airway patency, i.e., a neuromuscular compensation. In the tensor palatine muscle, an augmented muscle activity was seen in awake patients with OSA. However, during sleep, the decrements in activity were greater than in controls, which may represent a loss of the compensatory mechanism that is present during wakefulness (28). The palatopharyngeus muscle in the posterior palatine arch is a levator, and may as well be active in the efforts to maintain upper airway patency, as demonstrated in sleep apneics in a study of the palatoglossus in the anterior palatine arch (29). Abnormalities in upper airway muscles, as demonstrated in the palatopharyngeus in the present study, could explain such loss of compensatory mechanism during sleep.

Our findings of a local neurogenic lesion accord with those of a previous study of the soft palate (8), which is a structure even more subjected to the snoring trauma. In that study, atrophied and hypertrophied muscle fibers were demonstrated in sleep apneics and severe snorers, in addition to electronmicroscopic findings of frequent focal degeneration of myelinated nerve fibers in sleep apneics. In another study of bulldogs with OSA, moth-eaten fibers and other signs of morphologically abnormal muscle fibers were demonstrated in a pharyngeal dilator (30).

Several studies indicate that airway patency depends on a reflexogenic mechanism of afferent and efferent nerve pathways in the mucosa and upper airway muscles (5, 6). We suggest that a lesion of these nerves impairs the ability of reflexogenic dilation, leading to an increased upper airway collapsibility. A reduced pharyngeal lumen, because of this and/or other risk factors (obesity, anatomical abnormalities), could in vulnerable patients start a vicious circle, leading to an increased degree of periodic obstructive breathing causing more morphological abnormalities and neurogenic lesions, as indicated in the present study. Finally, a total collapse of the upper airway during inspiration and sleep may occur, as previously demonstrated in severe OSA patients.

In summary, the significant correlation between the time of sleep with periodic obstructive breathing and the degree of morphological abnormality, including specific neurogenic signs, supports the hypothesis of a progressive nature of snorers disease. It is suggested that the mechanical trauma of heavy snoring can explain these findings. As signs of a neurogenic lesion are seen already in several specimens of "simple snorers," effective prevention and/or treatment against snoring is implied.