INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nasal continuous positive airway pressure (nCPAP), which acts as a pneumatic splint within the upper airway (UA) (1), is the recommended therapy for patients with moderate to severe obstructive sleep apnea syndrome (OSAS) (2). The adequate determination of the efficient pressure requires in theory a full-night polysomnography. However, split-nights and autoCPAP titration commonly replace the formal CPAP titration and these procedures may alter the effectiveness of the treatment. For these reasons, long-term efficacy of nCPAP remains questionable in some subgroups of patients. Apart from some side effects related to nCPAP therapy (3), the main limitation remains that the device must be used every night for life (4). Such a constraint can be unacceptable in younger patients with OSAS with a prospect of long-term use over many years. Thus, surgery may be considered for obstructive sleep apnea in patients in whom other noninvasive treatments were unsuccessful or have been rejected on a long-term basis and who desire surgery and are medically stable enough to undergo the procedure (2).

Surgical intervention for snoring and OSAS includes several procedures, each designed to increase the patency of the upper airway at a specific level. The objectives of uvulopalatopharyngoplasty (UPPP) are to enlarge the oropharyngeal airway. The results of UPPP have been nicely summarized in a recent meta-analysis done by Sher and coworkers (5). A good response was defined as a 50% drop in the apnea plus hypopnea index (AHI) and the consequent achievement of an apnea index (AI) value of < 10 or an AHI of < 20. When these last criteria were taken into account (5) the response rate in the meta-analysis attained 41% (137 of 337 patients). Nonresponders had a higher baseline AI and AHI. A second step of the meta-analysis addressed the studies that included techniques aimed at determining the pharyngeal collapse site (9 studies, n = 168 patients). The percentage of patients who reached a 50% decrease in AHI and a postoperative AHI < 20 was much higher in patients demonstrating preoperative retropalatal collapse or narrowing than in patients with retrolingual collapse (46% versus 5%, respectively). Thus, when an imaging or endoscopic technique has undoubtedly shown an awake retrolingual narrowing or, more so, a retrolingual collapse during apneas, UPPP should be rejected. In patients with suspected hypopharyngeal collapse, maxillofacial surgery could be proposed if a surgical treatment is chosen.

There are two different treatment philosophies regarding maxillofacial surgery for OSAS. The Stanford group (6) has designed phased procedures tailored to the specific anatomical abnormalities encountered in each patient. The goal is to avoid a full maxillomandibular advancement osteotomy (MMO), at least in a number of patients. Phase 1 consists of limited mandibular osteotomy (with or without hyoid myotomy and hyothyroidopexy and with or without UPPP). In this procedure MMO is performed as the second or third step (phase 2 surgery). Conversely, other groups proceed (7, 8) directly to MMO. Palatal surgery is not systematically included in MMO at the first step (8). In cases of failure secondary refinements are tried by palatal advancement and/or velopharyngeal soft tissue corrections (8).

Maxillomandibular advancement osteotomy has previously been shown to be effective surgical technique for the treatment of severe sleep apnea. The results are reproducible in different centers (6, 8) and appear sustained on a long-term basis (9). The rationale for using phase 1 surgery is to avoid more radical and high-risk surgery if unnecessary, particularly in milder cases such as simple snoring, upper airway resistance syndrome (UARS), or moderate sleep apnea syndrome. Riley and coworkers reported an attractive success rate in phase 1 (61% of 249 patients) (6). The response rate was 75% in a study including 9 patients who underwent a partial genioglossal advancement and hyoid myotomy and hyothyroidopexy (GAHT) procedure (genioglossal advancement without hyoid myotomy and suspension) (10). Yao and coworkers (11) had a success rate of 68% with phase 1 (13 responders/19 patients). In contrast, two other studies had success rates for phase 1 of 42 and 17%, respectively (5 of 12 and 6 of 36 patients) (12, 13). Thus, phased surgery gives inconsistent results in different centers.

We have analyzed our prospective experience with 51 consecutive patients (64 surgical procedures) treated by the step-by-step maxillofacial surgery previously described by Riley and coworkers (6). Only 2 of the 53 patients initially treated were lost to follow-up.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Surgery was offered to symptomatic patients with moderate to severe sleep apnea syndrome for whom other noninvasive treatments were unsuccessful or had been rejected on a long-term basis, who desired surgery, and who were medically stable enough to undergo the procedure (2).

Preoperative Evaluation and Selection of the Patients

Before treatment a polysomnography was done to clearly characterize the disease and assess its severity. Patients with significant or unstable lung disease were excluded. All patients underwent an upper airway evaluation using physical examination and cephalometric roentgenograms. In these adult patients no preoperative orthodentistry was envisaged. All the patients addressed to maxillofacial surgery were nonsmokers or have quit smoking for at least 1 mo prior to surgery.

Polysomnography. Continuous recordings were taken of the electroencephalogram with electrode positions C3/A2-Cz/O1 of the International 10-20 Electrode Placement System, eye movements, chin electromyogram, and electrocardiogram. Airflow was measured by a pneumotachograph (Kontron Instruments, Saint-Quentin, France) signal or by oronasal thermistors and/or nasal cannula. Respiratory effort was evaluated by monitoring thoracic and abdominal movements and in some patients by measuring esophageal pressure by means of an esophageal catheter (Compliance catheter volgens E.K.G.S; International Medical, Zutphen, NL). Oxygen saturation was measured with a Biox-Ohmeda 3700 oximeter (Ohmeda, Louisville, CO).

The polysomnogram was scored manually according to standard criteria (14). Episodes of apnea were defined as complete cessation of airflow for 10 s or more. Hypopnea was defined as a greater than 30% decrease in oronasal airflow or amplitude of the thoracic and abdominal signals lasting for at least 10 s associated with an arousal and/or a desaturation  3%. We have chosen these hypopnea scoring criteria as they are widely used in the literature and have been recently proposed by the American Academy of Sleep Medicine Task Force (15). The threshold in oronasal flow decrement for scoring hypopnea was the same when using or not using a nasal cannula in association with thermistors. We scored microarousals using the ASDA criteria (16). Our definition for microarousal remained constant over time.

Apnea/hypopnea events were classified as central, obstructive, or mixed according to the absence or presence of breathing efforts. Esophageal pressure monitoring was used in patients who could tolerate a balloon.

Upper airway assessment. Physical examination (n = 53) was addressed to the three segments of the upper airway: rhino-, oro-, and hypopharynx. Examination of nasal cavity tried to identify septal deviation, turbinates hypertrophy, and nasal valve collapsus. A significant nasal obstruction was found in 7 of 45 of the phase 1 patients and led to an additional septoplasty during the phase 1 procedure. The length, width, and tone of the soft palate tone were systematically studied as the aspect of the posterior pharyngeal wall. A long redundant soft palate was present in all but four patients with prior UPPP. Hypertrophic tonsils were encountered in 40% of the surgical candidates. Hypopharyngeal obstruction from the base of tongue was clinically considered as significant in 80% of the patients. This was generally explained by small oral cavities with a normal tongue size. Considering cephalometry all the patients demonstrated hypopharyngeal narrowing. Finally, a clinical severe craniofacial dysmorphy was found in 7 of 53 patients. Of the patients 65% exhibited a mild retrognathia.

Lateral cephalometric radiographs were obtained (17). Briefly, patients were seated with their head in a neutral position with the gaze parallel to the floor and the teeth in occlusion. The X-ray plate was placed next to the left side of the face and the cone 4 m from the patient. Exposures were taken with the patient remaining still while slowly exhaling a moderately deep breath. Each study gave 4.7 mGy of radiation to the skin. We used the SNA (angle measurement from sella [S] to nasion [N] to subspinale [A]) and SNB (angle measurement from sella [S] to nasion [N] to supramentale [B]) angle on cephalometry to characterize the severity of craniofacial dysmorphy. SNA and SNB (in degrees) describe the maxillary and mandibular position in relation to the cranial base.

Selection of Patients and Choice of the Surgical Procedure

The factors favoring pharyngeal occlusion can be divided in two categories: abnormal airway structure and abnormal airway function. Surgical procedures are designed to enlarge and/or stabilize the upper airway. Anatomical abnormalities seem to play an important physiopathological role, particularly in lean patients. Thus, these subjects probably represent the best candidates for surgical therapy and we decided to exclude subjects with severe obesity unless a significant weight reduction could be obtained (see RESULTS).

The identification of collapsing or stenotic site(s) in the pharyngeal lumen seems critical in selecting the appropriate procedure. The efficacy of the methods for detecting the site of upper airway collapse has so far been less than desired and clear selection of good surgical candidates remains elusive. However, the likelihood of response of phase 1 tends to diminish in the more severely mandibular-deficient patients (6). Taking these data into account and using both the upper airway examination and the cephalometry, we have separated our surgical candidates in two groups depending on their degree of craniofacial dysmorphy.

Group 1. Patients demonstrating moderate craniofacial dysmorphy (SNB angle  75°) in whom pharyngeal narrowing was usually found in combination with moderate retrognathia and/or a long face appearance. A phased surgery was thus proposed to these patients (see following for details on surgical procedures).

Group 2. Patients demonstrating severe craniofacial dysmorphy (SNB angle < 75°) plus severe OSAS proceeded directly to phase 2. The surgical treatment was then specifically addressed to the craniofacial abnormalities and was usually a full maxillomandibular advancement osteotomy (MMO).

Surgical Techniques and Perioperative Management

Nasal surgery. If patients had obstructive nasal deformities, nasal reconstruction and/or reduction of turbinates were performed during phase 1 surgery or Le Fort 1 osteotomy.

UPPP techniquetonsillectomy. The steps for UPPP were classical and were done as follows: (1) tonsillectomy was performed if not previously done, (2) the anterior and posterior pillars were trimmed and reoriented, and (3) the uvula and the posterior portion of the palate were resected.

Phase 1. Phase 1 surgery was considered a conservative approach. Patients underwent a conservative tongue advancement, which consists in anterior repositioning of the genioglossus muscle and the hyoid bone (i.e., genioglossal advancement and hyoid myotomy and hyothyroidopexy: GAHT). As all the patients demonstrated both palatal and base of tongue narrowing, they underwent both GAHT and UPPP. The genioglossal advancement was initially performed using an inferior sagittal osteotomy (phase 1a) (see Figure 1A). The osteotomy could also begin at the inferior border of the mandible and was then followed under visual control by V-shaped osteotomy for advancement of the genioglossus muscle (phase 1b) (see Figure 1B). As later described we performed the stabilization of the hyoid bone inferiorly by attachment to the superior border of the thyroid cartilage (18). Six months after the operation, a polysomnographic reevaluation was done to determine improvement. Patients with significant residual sleep apnea could be offered phase 2 therapy.

View larger version (32K): [in this window] [in a new window]   Figure 1.   Genioglossal advancement, hyoid myotomy, and hyothyroidopexy (GAHT). (A) The genioglossal advancement is usually performed using an inferior rectangular sagittal osteotomy. (B) The osteotomy can also begin at the inferior border of the mandible and is then followed under visual control by trapezoidal osteotomy for advancement of the genioglossus muscle. The late modifications involve stabilization of the hyoid bone inferiorly by attachment to the superior border of the thyroid cartilage (16) (A and B).

Phase 2. Surgical treatment consisted of mandibular advancement by bilateral sagittal split osteotomy and maxillary advancement by Le Fort I osteotomy. Fixation in the new position was achieved by miniplates in the maxilla and bicortical miniscrews in the mandible (Figure 2). The maxillary advancement was at least 10 mm and frequently 12 mm targeting to a chin advancement of 15 mm (Figure 3).

View larger version (25K): [in this window] [in a new window]   Figure 2.   Maxillomandibular advancement osteotomy (MMO). Bilateral sagittal split osteotomy and maxillary advancement by Le Fort I osteotomy.

View larger version (59K): [in this window] [in a new window]   Figure 3.   Cephalometry before surgery (A), after phase 1 (B) and after MMO (C ). (A) Cephalometry before surgery. (B) Cephalometry after GAHT: the osteotomy of the mandible is visible. The hyoid bone is placed anteriorly and inferiorly by attachment to the superior border of the thyroid cartilage. Note that no enlargement of the upper airway occurred. The patient was not cured. (C ) After MMO, bilateral retromolar sagittal split osteotomy and maxillary advancement by Le Fort I osteotomy. Fixation in the new position is achieved by miniplates in the maxilla and bicortical miniscrews in the mandible (see the cephalometric picture). Note the significant increase in UA size. There was a normalization of the AHI after this procedure.

Perioperative Management

Anesthetic procedures and perioperative protocols were standardized. For all patients nCPAP was started at least 1 mo before surgery to stabilize their cardiovascular state and reduce UA edema. Patient compliance and tolerance were systematically verified. The patients were questioned regarding previous intubation difficulties and a fiberoptic intubation was performed when necessary. After surgery, extubation was done by the operating room staff, the patients being fully awake. Transfer to the intensive care unit (ICU) then occurred and the patients were systematically placed under nCPAP after phase 1 but not after phase 2. Patients were informed that during a 24-h period after surgery they would be kept in the ICU. Sedative and opioid premedications were omitted as the intra- and postoperative use of opioid was limited or avoided. All anesthetic drugs were administered by cautious titration to desirable effect, preferably using short-acting drugs. When feasible nonopioid analgesics were used for postoperative analgesia. Nocturnal oximetry was monitored throughout the hospital stay. Patients were asked to continue to use nCPAP after phase 1 until a polysomnographic recording had verified that surgery was a success.

Postoperative Evaluation

Patients were prospectively studied using polysomnography 6 mo after each step of maxillofacial surgery to review AHI and sleep parameters. Surgery was considered a success if the postoperative AHI was less than 15/h with at least a 50% reduction.

Statistical Analysis

A paired two-sample t test was used to evaluate the changes in polysomnography before and after surgery. Different subgroups of patients were compared using the Mann-Whitney or the two-sample t test. The Chi-square test was used for comparison of qualitative variables. All results are shown as mean ± SD. Significance was accepted for p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Fifty-one consecutive white patients (64 surgical procedures) operated between May 1994 and December 1997 were included in the surgical evaluation (Figure 4). During the same time interval nCPAP was started in 939 OSAS subjects. Forty-four patients have undergone phase 1 surgery. Twenty patients have had the benefit of maxillomandibular advancement (phase 2) (13 failures of phase 1, 7 directly the result of severe craniofacial dysmorphy [SNB angle < 75°]). Only 2 of the 53 patients initially treated were lost for follow-up. One of these two patients did not complete postoperative follow-up evaluation. This patient with a previously 5 yr stable pulmonary hemosiderosis had a deterioration of his lung condition (adult respiratory distress syndrome) immediately after the operation.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Enrollment of patients to the different surgical techniques and additional treatments. 1*: One patient did not complete postoperative evaluation because adult respiratory distress syndrome occurred after the operation. 1**: one patient lost for follow-up. Only 13 of 18 subjects who had MMO after phase 1 surgery failure have been included in the 6-mo evaluation of the present study.

Phase 1 Surgery (n = 44)

The first 21 patients were operated using the phase 1a surgical technique (associated nasal surgery: n = 2) (see Figure 1A). The 23 following patients were treated using the phase 1b surgical protocol (associated nasal surgery: n = 5) (see Figure 1B). The anthropometric and polysomnographic characteristics of the patients are summarized in Table 1. The population as a whole demonstrated moderate to severe sleep apnea syndrome (mean AHI = 45 ± 27/h). The majority of patients were moderately overweight (mean body mass index [BMI]: at entry = 26.3 ± 2.9 kg/m² [median = 26.9 kg/m²; range = 19.6-33.1]; at follow-up: 25.8 ± 2.9 kg/m² [median = 26.1 kg/ m²; range = 20-32.2], p < 0.01) middle-aged males. In our series, only three patients with a BMI over 30 kg/m² were operated and all three were surgical failures. The patients treated using phase 1b surgery were significantly less overweight and, as a consequence, although their mean diagnosis AHI was no different compared to phase 1a patients their cumulative time spent below 90% of Sa_O2 during the night was significantly lower (Table 1).

Efficacy of phase 1a surgery. The results of phase 1a were mediocre with 5 good responders (24%) and 16 poor responders (76%) as defined by polysomnographic criteria (postoperative AHI less than 15/h with at least a 50% reduction). The AHI for the group as a whole was not changed by phase 1a surgery (Table 2). There was a trend to a decrease in the apnea index with a significant increase in the hypopnea index, thus explaining the unchanged AHI. In spite of the absence of improvement in the number of microarousals, the sleep macrostructure improved with a significant increase in stages 3-4 and rapid eye movement (REM) sleep and a reduction in stages 1-2. The oximetric parameters were slightly improved with a significant increase in minimal nocturnal Sa_O2. The BMI for the group as a whole slightly decreased during phase 1a surgery (mean BMI: at entry = 27.4 ± 2.5 kg/m² [median = 27.2 kg/m²]; at follow-up: 27.0 ± 2.4 kg/m² [median = 26.9kg/ m²], p < 0.03).

Efficacy of phase 1b surgery. The success rate of phase 1b was only 22% (5 of 23 patients) (Table 3). The AHI for the group as a whole was unchanged after phase 1b surgery. All the oximetric parameters (mean, minimal Sa_O2 and CT < 90%) significantly improved after phase 1b. The BMI for the group as a whole slightly decreased during phase 1b surgery (mean BMI: at entry = 25.2 ± 3.0 kg/m² [median = 26.0 kg/m²]; at follow-up: 24.5 ± 2.8 kg/m² [median = 24.3 kg/m²], p < 0.01).

Overall efficacy of phase 1 surgery. The overall success rate of phase 1(a+b) was 22.7% with only 10 of the 44 patients cured (Table 4). The AHI for the group as a whole was unchanged by phase I surgery. There was a trend toward a decrease in the apnea index with a significant increase in the hypopnea index, thus explaining the stability of AHI. For the whole group, the sleep structure was modified with a significant reduction in stages 1 and 2 in favor of REM sleep. However, the number of microarousals was unchanged for the group as a whole. The minimal and the mean nocturnal Sa_O2 were significantly improved after the surgical procedure. Although there was a trend, we found no significant relationship between the success rate of surgery and the different classes of BMI: BMI < 25 kg/m², 27%; BMI 25-27 kg/m², 25%; BMI 27- 30 kg/m², 22%; BMI > 30 kg/m², 0% (n = 3). The mean genioglossus advancement for the entire group was 14 ± 2 mm.

Complications of phase 1. The usual hospital stay for phase 1 surgery was 5 d. Only one patient exhibited an acute upper airway obstruction because of a lingual hematoma. The patient was successfully managed without a long-term sequel with a surgical debridment performed under general anesthesia. All the patients complained of local oropharyngeal pain related to UPPP in the first 3-5 d after surgery. All patients experienced transient chin and lower lip anesthesia and/or paraesthesia. There was an 80% resolution between 3 and 12 mo. We did not observe mandibular fracture of the anterior mandibular fragment even when phase 1b was used. In three patients treated by phase 1a, the rectangular mandibular fragment resorbed, probably resetting the genioglossus muscle to its original position.

Additional Treatment Required after Phase 1

Thirty-four patients were considered as failure of phase 1. Additional treatment after maxillofacial surgery was required in 32 of these patients (94%). nCPAP was continued by 6 patients (17.6%) and refused after surgery by 4 patients (11.7%). Eighteen patients (52.9%) received phase 2 surgery (13 already evaluated by 6-mo polysomnography in the present study). Two patients (5.9%) received oral appliances. Finally, two patients (BMI = 27.8 and 25.5 kg/m²) demonstrating a persistent moderate sleep apnea after phase 1 (AHI = 21.3 and 11.9/h) were addressed for additional weight loss. In summary, additional treatment was required in 32 of 44 patients (72.7%) who underwent phase 1 surgery.

Phase 2 Surgery (n = 20)

Thirteen patients who failed phase 1 surgery elected phase 2 treatment. Seven other patients with severe OSAS proceeded directly to maxillomandibular advancement because of severe craniofacial dysmorphy (SNB angle < 75°). The respective characteristics of the two subgroups of patients are shown in Table 5. The mean advancement of both maxilla and mandible was 11.8 ± 0.5 mm. Although all patients lost between 4 and 15 kg initially after surgery, at the postoperative recording, the BMI had returned to baseline. The BMI for the group as a whole slightly decreased during phase 2 surgery (mean BMI: at entry = 26.9 ± 4.3 kg/m² [median = 27.0 kg/m²; range = 19.6-36.6]; at follow-up: 25.4 ± 3.3 kg/m² [median = 24.9kg/ m²], p < 0.01). Phase 2 was highly effective with a significant improvement in AHI, sleep, and oximetric parameters (Table 6). The success rate was 75% (15 of 20 patients). Three of the five patients considered as failures demonstrated an AHI less than 20/h with two of the three patients becoming asymptomatic. The results were not significantly different between the two subgroups of patients (i.e., failure of phase 1 or directly operated using MMO). There was no relationship between the success rate of surgery and the different classes of BMI: BMI < 25 kg/m², 71%; BMI 25-27 kg/m², 67%; BMI 27-30 kg/m², 83%; BMI > 30 kg/m², 75%.

Complications of phase 2. The usual hospital stay for phase 2 surgery was 7 d. Postoperative recovery after phase 2 surgery was difficult for the patients but there were no major complications and minor complications occurred only in three cases. In one patient there was a local infection that responded to antibiotics. One patient had a perforation of the palate during maxillary advancement, which healed spontaneously over 10 d. A third had a maxillary pseudarthrosis, which led to abnormal mobility, which appeared after 3 yr and required curettage and secondary bone graft. Technical difficulties arising from the phase 2 operation were related to the absence of preoperative orthodontistry, which is difficult in adults, the mediocre state of teeth and bone in these middle-aged patients, and the degree of maxillary advancement that required rigid osteosynthesis in order to achieve long-term skeletal stability. In general, recovery of full mandibular function was difficult with the frequent occurrence of hypoesthesia of the lower lip. All patients who had a prior phase 1 including UPPP had postoperative velar insufficiency of a temporary nature. This was predominantly a phonetic deficit without liquid regurgitation and was usually exaggerated by neck extension and required several weeks of speech therapy. The phonetic problems were initially aggravated by elocution difficulties related to the increased tension of the labial muscles related to dental advancement. This late condition was improved by cautious speech therapy. The aesthetic impact of maxillomandibular advancement on facial appearance was usually mainly represented by an enlargement of the nose base and nostrils associated with a convex aspect of the superior lip. No subsequent need for additional treatment occurred. The patients were informed, before surgery, of the modifications in facial appearance potentially induced by MMO. All the complications were increased with increasing age, in particular after 45 yr. The average time off-work postoperatively was 10 wk. There were no long-term problems in either speech or swallowing in any of the patients. Paradoxically, this step of the procedure led to less complaints of pain than phase 1 surgery or UPPP.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our study demonstrated a success rate of 22.7% for phase 1 surgery (i.e., limited mandibular osteotomy with hyoid myotomy, hyothyroidopexy, and UPPP) in a large consecutive group of 44 moderate to severe lean patients with OSAS. In 32 of the 44 patients who underwent phase 1 surgery (72.7%) additional treatment was required. Conversely, maxillomandibular advancement surgery (phase 2) was successful in 75% of 20 severe patients with OSAS.

Phase 1 Surgery

Comparison between our series and others. In 249 patients reported by Riley and coworkers the success rate of phase 1 (reduction of AHI by more than 50% and AHI less than 20/h) was 61% (6). The likelihood of response tended to diminish with increasing preoperative AHI severity, the nonresponders being more obese and more severely mandibular deficient than the responders. In this paper the results of different surgical techniques were reported together and 109 of the 415 patients (26%) did not receive postoperative polysomnography or were lost to follow-up. A study in 9 patients who underwent a partial GAHT procedure (genioglossal advancement without hyoid myotomy and hyothyroidopexy) reported a success rate of 75% (10). Yao and coworkers (11) found a success rate of 68% with phase 1 (13 responders/19 patients) but in this study only 19 of the 44 operated patients were finally included in the analysis. Unlike these relatively satisfactory results, two other studies found a success rate for phase 1 of, respectively, 42 and 17% (5 of 12 and 6 of 36 patients) (12, 13). We found a success rate for phase 1 of 22.7%. In our series, patients usually demonstrated a partial improvement with a trend toward a decrease in the apnea index and a significant increase in the hypopnea index thus explaining the stability of AHI. For the whole group, the sleep structure was modified with a significant reduction in stages 1 and 2 in favor of REM sleep. However, the number of microarousals was unchanged for the group as a whole. The minimal and the mean nocturnal Sa_O2 were significantly improved after the surgical procedure. Thus, everything indicates that such surgery was only partially able to prevent UA collapse with persistent sleep hypopneas and associated sleep fragmentation.

Factors of discrepancy between our results and previous more satisfactory experiences with phase 1. In our routine polygraphic recordings, we now consistently use nasal pressure measured by nasal prongs. This technique has the advantage of measuring airflow during sleep in a semiquantitative way (19, 20). On the other hand, decreases in nasal pressure can correspond to an apneic/hypopneic event but can also result in a shift from nasal to mouth breathing. Taking these technical limitations into account, we record oral and nasal thermistry, in a concomitant way, to assess oral and nasal flow as a sum. Thus, we are able to compare the two signals and identify false apneas on nasal pressure actually corresponding to hypopneas with preferential mouth breathing. This set of recording-associating thermistors and nasal pressure was used in the present study and provided a within-patient consistency. Indeed, even for patients initially diagnosed only with thermistors, the same kind of information was available after surgery.

When nasal pressure recordings are used to detect sleep- related breathing events the AHI is significantly higher than with thermistry alone or thermistry associated with inductive plethysmography (mean difference 4.5 events/h) (21). It can be argued that this has played a role in our study in increasing the number of poor responders to phase 1 surgery. However, there was no major discrepancy between thermistors and nasal pressure results that could even partly explain our results. Actually an expected mean difference of less than 5 event/h between the two diagnostic modalities would have been clearly insufficient to explain a residual AHI of 42.8/h after phase 1. Moreover, it should be noted that these residual events were biologically significant as associated with a desaturation and/ or a microarousal. Finally, we have used the same diagnostic techniques for phase 2 evaluation and in this subgroup of patients we have found results that were concordant with previously published studies.

The pathological significance of hypopneas is not definitely established. However, it is usual to include hypopneas in the polygraphic evaluation of surgical results. In the majority of the studies addressing phase 1 surgery (6, 8, 10) the hypopneas were determined using thermistors that cannot detect subtle changes of respiratory flow, which may result in a significant underestimation of residual hypopneas after surgery. This point is probably of great importance as we have demonstrated that in a majority of patients there was a transition from apneas to hypopneas, the later becoming the main type of residual respiratory events after surgery. Moreover none of the previous studies reported results of microarousal index before and after surgery (16). Similarly, the significant residual sleep fragmentation we have found after surgery was in accordance with our 77.3% rate of phase 1 failure.

The proportion of patients with postoperative polysomnography was very different from one study to another, varying between 43% (11) and 96% of the patients in the present study. Such huge differences in the rate of patients lost to follow-up are probably part of the explanation for the different published success rates.

The aim of phase 1 surgery is to advance the anterior mandible with the attached genial musculature to reduce retrolingual obstruction. The "window osteotomy" corresponding to phase 1a demonstrated the following problems (22): (1) poor visualization of the musculature to be advanced, (2) twisting of the segment along with the musculature through 90°, with potential risk of tendon rupture, and (3) difficulty in securing the advanced segment. Our first 21 patients were operated using the phase 1a surgical technique. After finding a low success rate with only five good responders (24%) we hypothesized that such results could be partly explained by the quality of the genioglossal advancement. Thus, the 23 subsequent patients were treated using the phase 1b surgical protocol that allowed a more selective genioglossal advancement under visual control without involving the geniohyoid muscles. The mobilized bone fragment was bigger using phase 1b surgery to reduce bone resorption and avoid torsion on the neurovascular bundle in the pedicle. Unfortunately this later surgical technique did not lead to a better success rate.

In summary, owing to our meticulous measurement of respiratory events and sleep fragmentation, our low rate of patients lost to follow-up and the reproducibility of the phase 1 results whatever the surgical technique used, we are confident of the 22.7% success rate we found for phase 1 surgery. Although this success rate could appear unsatisfactory, this much less traumatic treatment than phase 2 surgery might be a therapeutic solution for a carefully selected subgroup of patients.

Hypothesis regarding phase 1 surgery inefficacy. The factors favoring airway occlusion can be divided into two categories: abnormal airway size and abnormal airway function. In the literature, there are only a few studies focusing on the caliber of the upper airway using cephalometry before and after phase 1 surgery (11, 23). Riley and coworkers (24) found that poor responders were patients more mandibularly deficient before surgery or with a lesser amount of advancement of genioglossus muscle after surgery (15 versus 12 mm in success and failure of phase 1, respectively). These data suggested that in phase 1 failure, the remaining collapse during sleep after surgery occurred mainly at the hypopharyngeal level. However, numerous studies (25) have demonstrated that in UPPP failures (a part of the phase 1 surgical procedure) the collapse of the upper airway remained at the level of the oropharynx as a result of the persistence of an oropharyngeal narrowing after UPPP associated with an increase in the thickness of the soft palate after UPPP (28). Thus, further studies are needed including three-dimensional imaging techniques to better describe the changes in upper airway dimensions before and after surgery at the different pharyngeal levels.

Abnormalities in pharyngeal function include abnormal function of the pharyngeal dilator muscles and increased collapsibility of the pharyngeal walls. Pharyngeal collapsibility can be estimated using the measurement of pharyngeal critical pressure (Pcrit) (29). Schwartz and coworkers (30) looking at 13 patients have reported that the response to UPPP is determined by the magnitude of the fall in the upper airway Pcrit, i.e., the amplitude of reduction in pharyngeal collapsibility. A measurement of pharyngeal collapsibility (Pcrit) before surgery is probably of interest. To date no data are available on the effects of phase 1 surgical techniques on pharyngeal dilator muscle activity, contraction efficacy, and upper airway collapsibility.

Follow-up and additional treatments required after phase 1 surgery. In our study 73% of the patients required additional treatment after phase 1. This confirms the importance of routine polysomnography postoperatively. Follow-up should never consist solely of interviewing the patient regarding his or her symptoms. Patients should be informed prior to surgery that the surgery may fail to treat their condition adequately and that they may need additional treatment postoperatively. However, it should be emphasized that 4 of 10 patients in whom nCPAP was indicated refused the device after surgery. The explanations for that could be a loss of confidence in care induced by the failure of surgery, an insufficient management of nasal problems, or an increase in mouth leaks. The operation may have compromised the seal between the soft palate, tongue, and palatal arches. One study has shown that UPPP may compromise nCPAP therapy by increasing mouth leaks and reducing the maximum level of pressure that can be tolerated (31). The frequent occurrence of such a secondary refusal to pursue nCPAP in surgical failures should be taken into account when phase 1 is envisaged as a therapy for sleep apnea. Further studies should specifically be addressed to nCPAP compliance after surgical failures.

A rationale for continuing to use phase 1 surgery despite the 22.7% success rate might be based on the risk-benefit ratio. Maxillomandibular advancement osteotomy with its potential morbidity may not be justified for simple snoring, UARS, and even moderate sleep apnea syndrome. In these subgroups of patients, however, oral appliances or conservative treatments including weight loss, change in sleep position, and avoidance of evening alcohol may be an alternative.

Phase 2 Surgery

Maxillomandibular advancement osteotomy is an effective surgical technique for the treatment of severe sleep apnea. The results are reproducible in different centers (6, 8) and appear sustained on a long-term basis (9). Hochban and coworkers (8) did not perform phase 1 prior maxillomandibular advancement osteotomy and obtained a success rate of 90%. This is in contradiction with the results of Waite and coworkers (7) who found a good response to surgery only if at least an adjunctive procedure (phase 1) was done in addition to MMO. The Stanford group also stated that staged changes in the position of bony structures (i.e., step-by-step surgery) could lead to better sustained muscle elasticity than repositioning the structures in one large step. This hypothesis is purely speculative and muscle activity measurements and an assessment of UA collapsibility are needed to demonstrate this assertion. The results of Hochban and coworkers (8) and our own results demonstrate that some moderately overweight but skeletally deficient patients may be appropriately treated by phase 2 surgery without undergoing the phase 1 procedure. These patients when young and motivated may represent the only surgical indications in severe OSAS. In our experience this subgroup represents less than 2.5% of the patients recruited in a university sleep laboratory (20 phase 2 versus 939 nCPAP during the time of the study).

Conclusion

Phase 1 surgery cured 22.7% of our patients. A rationale for continuing to use phase 1 despite this low success rate might be based on the risk-benefit ratio. Phase 1, a much less traumatic technique than phase 2, is effectively able to treat a subgroup of OSAS patients. The results of phase 2 surgery are successful in young patients with severe OSAS even if the surgical technique is more aggressive. Formal assessment of benefits including quality of life measurements are required to compare such a therapy with long-term nasal CPAP. A long-term follow-up is also needed to establish whether these results are stable on a long-term basis.