This clinical randomized controlled trial was performed to compare the effects of distraction osteogenesis (DO) and conventional orthognathic surgery (CO) on velopharyngeal function and speech outcomes in cleft lip and palate (CLP) patients. Twenty-one CLP patients who required maxillary advancement ranging from 4 to 10 mm were recruited and randomly assigned to either CO or DO. Evaluation of resonance and nasal emission, nasoendoscopic velopharyngeal assessment and nasometry were performed preoperatively and at a minimum of two postoperative times: 3–8 months (mean 4 months) and 12–29 months (mean 17 months). Results showed no significant differences in speech and velopharyngeal function changes between the two groups. No correlation was found between the amount of advancement and the outcome measures. It was concluded that DO has no advantage over CO for the purpose of preventing velopharyngeal incompetence and speech disturbance in moderate cleft maxillary advancement.
Maxillary hypoplasia is a common developmental problem in individuals with cleft lip and palate (CLP) and is thought to result from a combination of a congenital reduction in midfacial growth and the effects of the surgical scarring from cleft palate repair . Many patients with this problem can benefit aesthetically and functionally from surgical correction. Traditionally, maxillary advancement was corrected by what is now termed conventional orthognathic surgery (CO) although, since 1997, distraction osteogenesis (DO) has become an alternative option for the treatment of maxillary hypoplasia in patients with CLP . When the maxilla is advanced surgically, the soft palate is thought to move forward, which may affect velopharyngeal function (VF). At the same time, improved occlusal relationships may improve speech articulation. Studies reporting the influence of maxillary advancement on speech and VF have yielded different results. In a critical review of the effect of cranio-maxillofacial osteotomies and DO on speech and velopharyngeal status conducted by Chanchareonsook et al. , the results varied considerably. While many studies reported that surgical advancement of the maxilla had no impact on speech and velopharyngeal status, others reported worsening in patients with pre-existing velopharyngeal impairment or those with borderline VF before surgery . With regards to articulation, some studies found improvement in articulation and some no change . Chanchareonsook et al. also noted that there were no clear differences in outcomes between CO and DO, although there were few systematic comparisons. The authors’ centre has been conducting a prospective randomized controlled trial to compare the speech outcomes and velopharyngeal status of patients with repaired cleft lip and palate, who underwent either CO or DO to correct maxillary hypoplasia. The early results (3 months postoperatively) showed that there was no statistically significant difference in either speech or VF between the two surgical techniques . The sample size was small and the follow-up time short, so a longer follow-up with a larger sample size was recommended .
This study aimed to compare the long-term effects on speech (resonance and nasal emission) and VF in two groups of CLP patients indicated for moderate advancement, one treated with DO and the other with CO.
The null hypothesis has been adopted to avoid bias: that there is no significant difference in the long-term effects on speech and VF in CLP patients undergoing either CO or DO.
Materials and methods
All patients with CLP, who were over 16 years of age and skeletally mature, and presenting with moderate maxillary hypoplasia requiring Le Fort I osteotomy were evaluated for recruitment into this study. The maxillary alveolar clefts were grafted and pre-surgical arch alignment and decompensation by fixed orthodontics completed. This was followed by a standard surgical assessment, dental model surgery and prediction tracing. Only patients who required maxillary advancement ranging from 4 to 10 mm based on the model surgery were recruited into the trial. Others with smaller planned advancement underwent CO, whereas those with larger advancement underwent DO and were excluded from the trial. Syndromic patients and patients who presented with systemic diseases were excluded. Patients who satisfied the inclusion criteria were randomly assigned into two surgical groups, for treatment by DO with internal maxillary distractors or by CO with mini-plate fixation.
A senior investigator (LKC) with extensive experience in both distraction and orthognathic surgery performed the clinical assessment and was in charge of the randomization of the patients using a table of random numbers generated by computer. The allocation was concealed until informed consent was being explained to the patients.
The study was approved by the Faculty of Dentistry Ethics Committee and all patients involved in the study provided written consent. The surgery was supervised by LKC and was conducted in the Discipline of Oral and Maxillofacial Surgery, The University of Hong Kong between June 2002 and January 2008.
A standard Le Fort I osteotomy and down fracturing was performed . Maxillary segmentalization was carried out if necessary according to the dental model surgical plan.
In the CO group, the maxilla was fully mobilized to the pre-planned position. Custom-made arch bars were ligated to the maxillary and mandibular dental arches and the teeth placed in the occlusal wafer followed by intermaxillary fixation with wire loops. The mobilized maxilla was fixed with two titanium mini-plates on each side at the zygomatic buttress and the pyriform region. Intermaxillary fixation was removed and the accuracy of the dental interdigitation was checked.
A standardized technique of maxillary distraction with the use of internal distractors was developed for the DO group. A similar vestibular incision and bone cuts as for CO was made. The maxilla was fully mobilized but not transposed to the final occlusal position. Internal bone-borne maxillary distractors (Synthes, Pennsylvania, USA) were then inserted. The distractors on either side were activated for a few millimetres to check the correctness of maxillary transport, with the aim of reaching good occlusal interdigitation. The mucosal wound was then closed by continuous suturing to leave the activator rod external to the mucosal wound for later activation. Mandibular osteotomies, if needed, were performed in the same operation.
After a latency of 3 days, activation was commenced at 1 mm per day in two rhythms with an activating key by the patients themselves or their relatives until a class I incisal relationship was achieved. Light orthodontic elastic traction was applied to control the occlusion in both groups during the early postoperative period.
The primary outcome of the study was the long-term skeletal stability of CO and DO in CLP patients. This was assessed by comparing serial lateral cephalographs taken at different postoperative times.
The secondary outcomes of the study were dental relapse, morbidities, speech and VF changes, soft tissue changes and psychological profile changes. Clinical relapse and morbidities were recorded using questionnaires. Speech and VF changes were assessed using nasoendoscopy, perceptual speech assessment and nasometer. The changes in the soft tissue were compared using serial lateral cephalographs taken preoperatively and postoperatively at different time intervals. Psychological profiles were measured using standardized questionnaires (Social Avoidance and Distress Scale, Satisfaction with Life Scale and Culture-free Self Esteem Inventory). This manuscript will only focus on two of the secondary outcomes: speech and VF changes.
The sample size was calculated based on the primary outcome. In the pilot study, the skeletal relapse of cleft maxillary distraction and orthognathic surgery was compared. A sample size of 42 ( n = 21 per group) was suitable for obtaining statistically valid data. The authors were able to determine a statistically significant difference ( p < 0.05) between CO and DO when the mean horizontal relapse was more than 0.14 mm at 80% power.
All patients underwent a nasoendoscopy examination preoperatively and at 3, 12 and 24 months postoperatively to determine their velopharyngeal status. All patients were examined using a flexible nasoendoscope (Kay Rhino-Laryngeal Stroboscope RLS 9100B, KayPentax, Lincoln Park, NJ; Olympus ENF type P-4, Olympus America Inc., Melville, NY, USA). The scope was connected to a cold light supply to allow visualization. The image obtained from the scope was visualized on a colour monitor and recorded onto the computer.
Patients were asked to sit with their necks extended and head tilted backward about 15°. Topical anaesthesia using 4% cocaine was sprayed into the nasal cavity in order to maximize patient comfort and cooperation and to diminish secretion. The nasoendoscopy was performed using the same nostril in each subject at the preoperative and postoperative examinations by the same senior investigator (NS).
After introduction of the nasoendoscope through the nostril, the tip was positioned above the level of the velopharyngeal orifice. The boundary of the velopharyngeal orifice was defined by the posterior border of the soft palate (velum), left and right lateral pharyngeal walls and the posterior pharyngeal wall. Each patient was asked to provide a speech sample, as instructed by the attending speech-language pathologist (TW). The ‘Cantonese Nasendoscopy Speech Protocol’, developed by Whitehill was used. The protocol asks patients to pronounce their name, age, count from 1 to 20, and requires sustained production of two vowels, consonant-vowel repetitions (e.g. ‘pa pa pa’, ‘ti ti ti’), sustained/s/, sentences loaded with plosives, fricatives and nasals, and non-speech activities such as blowing, whistling and swallowing. Each assessment took about 3–5 min.
For patients presenting with pharyngeal flaps, both pharyngeal orifices were included in the assessment so that all borders could be seen in a single view. Each portal was examined individually.
The clinical nasoendoscopy findings were evaluated on completion of each examination and the diagnosis was noted in the clinical record. This included: classification of VF (adequate closure, borderline competence, mild, moderate or severe incompetence), the consistency of the VF, any phoneme effects (e.g. better closure achieved for plosives), and pattern of closure (e.g. circular, coronal) . These clinical observations were not used for assessment of VF in this study. Rather, all the nasoendoscopy data were retrieved and transferred to a mini-digital video tape (Sony Corporation, Tokyo, Japan) by a technician (SH) not involved in the study. The videos were edited by this technician (SH) using Music Fan’s Factory (Color7 Tech, Alabama, USA) to eliminate the patient’s name and age. The videos were randomized across groups (CO and DO) and the time of evaluation (preoperative and postoperative) and were assigned a number by the technician (SH).
The analysis design was a randomized blind evaluation of the velopharyngeal gap closure. The 6-point rating scale developed by Chanchareonsook et al. was used, with the velopharyngeal resting position as the baseline ( Table 1 ). The scale was based on the protocol recommended by Golding-Kushner et al.
|0.00 =||Severe VPI, or no movement of velopharyngeal structures|
|0.25 =||Severe VPI, or minimal velopharyngeal closure|
|0.50 =||Moderate VPI, or moderate velopharyngeal closure|
|0.75 =||Mild VPI, or velopharyngeal closure between moderate and touch closure|
|0.90 =||Borderline touch closure, or pin hole closure|
|1.00 =||No VPI, or complete velopharyngeal closure|
|VPI = velopharyngeal incompetence|
The rating procedures followed those developed by Chanchareonsook et al. namely: a panel of three judges (HDC, TW and NS) analyzed the data by examining velopharyngeal movement during the nasoendoscopic procedure. A judgment was made independently by each examiner after each sample. A comparison of the ratings between the examiners was then made. If all three or two of the three assessors provided the same rating, this represented the final rating for any specific sample. If the three assessors had three different ratings, the rating was repeated until there was agreement between at least two judges; discussion was allowed during such cases. If the pattern of velopharyngeal closure was inconsistent, the maximum velopharyngeal closure during the entire protocol was used to represent the final rating. In cases where pharyngeal flaps were present, the judges were asked to focus on the port with the poorest closure.
Perceptual speech assessment
Speech examinations were carried out on the same visit as the velopharyngeal assessment. During each session, an experienced speech-language pathologist examined each patient to evaluate resonance (hypernasality and hyponasality), nasal emission and articulation. The articulation results are not reported here, but have been analyzed separately.
The speech examination was conducted in a quiet room. Each patient was asked to read a standard Cantonese/Chinese reading passage and the Cantonese Osteotomy Deep Test, a list of single words designed to tap the Cantonese phonemes most vulnerable to change with osteotomy . Speech samples were recorded using a Sony TCD-D3 Digital (DAT) tape recorder. A Sony ECM-909 microphone was used, with a mouth to microphone distance maintained at approximately 10 cm.
Evaluation of hypernasality, hyponasality and nasal emission was based on a 30 s sample edited from each reading passage speech sample using Main Actor V5.5 (MainConcept, Aachen, Germany) by a technician not involved in the study. The audiorecordings were re-named as numbers, randomized (across surgical groups and time of evaluation) and transferred to a compact disc, which was played using a Mac mini-computer (Apple, California, USA). The parameters involved in the speech analysis and the rating scales are presented in Table 2 .
|Speech variable||Rating scale|
|Hypernasality||0 = No hypernasality|
|1 = Mild hypernasality|
|2 = Moderate hypernasality|
|3 = Severe hypernasality|
|Hyponasality||−1 = Hyponasality|
|0 = No hyponasality|
|Nasal emission||1 = Present|
|0 = Absent|
Consensus listening was undertaken by two qualified speech-language pathologists (TW, JC), both of whom were experienced in evaluating the speech of patients with cleft palate. Both (TW, JC) were blinded to the patient group and whether the samples were preoperative or postoperative. The raters listened to the audio-samples at the same time using Sennheiser HD 212 Pro headphones (Sennheiser Electronic Corporation, Connecticut, USA). They made their judgments independently and then compared their findings. If there was a disagreement, the samples were played again and discussed until a consensus was reached. After reaching a consensus, the rating was recorded.
Nasality was assessed using a nasometer (Kay Elemetrics, Model 6200, KayPENTAX, New Jersey, USA). The nasometer was calibrated according to the manufacturer’s instructions, and the headset device was adjusted by the speech-language pathologist. This headset device separated the oral and nasal cavities using a baffle plate. Two microphones mounted on the top and bottom of the plate collected acoustic energy during speech. The instrument then computed a ratio of the acoustic data acquired by the two microphones (nasal energy divided by oral plus nasal energy). This ratio is called nasalance, which is considered an acoustic correlate of perceived nasality and is expressed as a percentage. Calibration was made before the assessment of each subject. Each patient read an oral passage aloud , which contained no nasal phonemes. If an error occurred during reading, the subject was asked to repeat the test.
The mean percentage nasalance was calculated for each patient at each time period. Previous studies have recommended various cut-off points for nasalance values such as 32% and 26% . In this study, a nasalance value higher than 30% was considered to be indicative of hypernasality . A 2-point nasalance scale was thus derived, where 1 is above/outside normal limits (nasalance above 30%), 0 is within normal limits (30% nasalance or below).
In view of missing data at various postoperative time points, only two postoperative data points for each patient were included in all the analyses. Wherever available, the third postoperative data was chosen. This resulted in a postoperative mean time of 4 months (range 3–8 months; SD 4.13) and another of 17 months (range 12–29 months; SD 5.78). These first and second postoperative time points will be referred to as post1 and post2, respectively.
An exact χ 2 test was performed to determine if there was a statistically significant difference between patients who showed an improvement, no change or deterioration in the different assessment parameters. Pearson’s correlation was performed to determine any correlation between the amount of surgical maxillary advancement and the different assessment parameters.
Twenty-one patients (11 CO, 10 DO) had a complete data set and were included in the nasoendoscopic assessment ( Fig. 1 ). The demographic profiles of the patients are presented in Table 3 . For these 21 patients, 63 samples (video image plus speech recording), separated into preoperative, post1 and post2, were available for assessment. Three samples were randomly selected, re-named and used for the intra-judge (test–retest) reliability. These 3 were added to the original 63 samples, forming a total of 66 samples. The 3 assessors were in agreement in 54 (81%) samples. In the remaining 12 samples (19%), the assessors were in partial agreement (i.e. 2 of the 3 assessors gave the same rating). For those samples where there was partial agreement, it was noted that there was only one scale difference in the ratings given. For the three randomly selected samples for the intra-judge reliability, all three assessors gave the same rating as for the original samples (100% intra-judge reliability).
|CO group||DO group|
The nasoendoscopy ratings for individual patients are presented in Table 4 . At the preoperative stage, 5 patients had borderline touch closure (0.9), 2 had mild velopharyngeal incompetence (VPI; 0.75), 1 moderate VPI (0.5) and 13 had complete velopharyngeal closure (1.0). At post1, 6 patients presented with borderline touch closure, 1 patient moderate VPI and 14 complete closure. At post2, 7 patients had borderline touch closure, 1 moderate VPI and 13 had complete velopharyngeal closure.
|Patients||Amount of advancement (mm)||Nasoendoscopy rating|
Of patients who presented with incomplete velopharyngeal closure at post1 ( n = 7), 3 patients (C2, C15, C19) with borderline closure were from the CO group, with maxillary advancement of 4 mm (2 patients) and 9 mm (1 patient). The remaining four patients were from the DO group, three patients (D6, D14, D20) with maxillary advancement of 4, 5 and 6 mm were found to have borderline touch closure. One patient (D4), with an advancement of 7 mm had moderate VPI.
The nasoendoscopy results were further classified into patients who showed an improvement, no change, or deterioration in velopharyngeal status after surgery. An exact χ 2 test showed no significant difference between the number of subjects who showed ‘improvement’, ‘deterioration’, and ‘no change’ in their velopharyngeal status between the two types of surgery at post1 ( χ 2 = 0.56, p > 0.05) and post2 ( χ 2 = 0.49, p > 0.05). Pearson’s correlation also showed no significant correlation between the amount of maxillary advancement and the velopharyngeal status at post1 ( r = 0.04, p > 0.05) and post2 ( r = 0.08, p > 0.05).
Twenty-two patients (11 CO, 11 DO) with complete data sets were included in the resonance analysis. Three samples were randomly selected, re-named and used for the intra-judge reliability. These 3 samples were added to the original 66 samples making a total of 69 samples available for assessment. The samples were randomized across surgical groups and times.
For the 69 samples, 65 (94%) rating judgments were agreed by the 2 assessors. The remaining four samples (6%) were played again and discussed in order to arrive at the final judgment. For the three randomly selected samples used for the intra-judge reliability, both assessors gave the same rating when compared with the original samples (i.e. 100% agreement).
In the CO group, the resonance of all 11 patients was considered normal preoperatively ( Table 5 ). At post1, four patients were rated as mildly hypernasal, and their status remained as such at post2. Only one patient (C11) showed hyponasality at post1. Preoperatively and at post2, his resonance was judged to be within normal limits. This patient was excluded from further analysis as the hyponasality was related to a cold or allergic rhinitis on the day of examination.
|Patient||Amount of advancement (mm)||Hypernasality||Nasal emission|