Predicted probabilities of surgical success among patients who had a tracheotomy prior to mandibular distraction osteogenesis. A, Females; B, males. The colored dotted lines indicate differences in the probability of success for each diagnosis. Open circles represent the observed values.
The predicted probabilities of complications stratified by the initial surgical intervention. A, Mandibular distraction osteogenesis (MDO) first; B, tracheotomy first. The colored lines indicate the differences in the probability of success for patients who had less than 2 vs 2 or more other airway procedures. The shaded areas represent the 95% CIs for these probabilities. Open circles represent the observed values.
The flowchart demonstrates our current algorithm for workup and decision making regarding the choice of mandibular distraction osteogenesis (MDO) vs tracheotomy for symptomatic micrognathia. CFM indicates craniofacial microsomia.
Lam DJ, Tabangin ME, Shikary TA, Uribe-Rivera A, Meinzen-Derr JK, de Alarcon A, Billmire DA, Gordon CB. Outcomes of Mandibular Distraction Osteogenesis in the Treatment of Severe Micrognathia. JAMA Otolaryngol Head Neck Surg. 2014;140(4):338-345. doi:10.1001/jamaoto.2014.16
Patients with severe micrognathia are predisposed to airway obstruction. Mandibular distraction osteogenesis (MDO) is an alternative to tracheotomy that lengthens the mandible in order to improve the retrolingual airway. This study presents outcomes from one of the largest cohorts reported.
To assess the rate and predictors of surgical success and complications among (1) patients who underwent MDO prior to other airway procedures (MDO first), and (2) patients who required an initial tracheotomy and were subsequently treated with MDO (tracheotomy first).
Design, Setting, and Participants
Retrospective cohort study at a tertiary care pediatric medical center of patients diagnosed as having micrognathia resulting in symptomatic airway obstruction (Pierre Robin sequence) and who underwent MDO from September 1995 to December 2009.
Electronic medical records were reviewed. Multivariable regression analysis was used to assess for predictors of outcome.
Main Outcomes and Measures
Rates of surgical success (defined as either tracheotomy avoidance or decannulation) and complications. Potential predictors included demographics, syndrome presence, follow-up time, and surgical history.
A total of 123 patients (61 in MDO-first subgroup, 62 in tracheotomy-first subgroup) underwent MDO during the study period. Median age at time of distraction was 21 months (range, 7 days–24 years). Surgical success and complication rates were 83.6% and 14.8% in the MDO-first subgroup and 67.7% and 38.7% in the tracheotomy-first subgroup. Tracheotomy-first patients were more likely to have a syndromic diagnosis (66.0% vs 43.0%; P = .009) and were older at the time of MDO (median age, 30 months vs 5.1 months; P < .001). Poorer odds of success were associated with the need for 2 or more other airway procedures (odds ratio [OR], 0.14 [95% CI, 0.02-0.82]) in the MDO-first subgroup and craniofacial microsomia or Goldenhar syndrome (OR, 0.07 [95% CI, 0.009-0.52]) in the tracheotomy-first subgroup.
Conclusions and Relevance
Mandibular distraction osteogenesis has a high rate of success in avoiding tracheotomy. Patients who required a tracheotomy before MDO had a lower success rate in achieving decannulation and a higher rate of complications. However, these patients also had a higher rate of syndromic diagnoses and associated comorbidities. Patients with Goldenhar syndrome have a decreased likelihood of surgical success.
Pierre Robin sequence is characterized by the triad of micrognathia, glossoptosis, and resultant airway obstruction owing to constriction of the retrolingual space.1 Though not classically described as part of its definition, cleft palate is a commonly associated finding that occurs in up to 90% of children with Pierre Robin sequence. Symptoms of airway obstruction may range from snoring and stertor while asleep to frank obstruction and retractions when awake. Gasping or aspiration while feeding is frequently encountered, and this in combination with chronic airway obstruction can cause failure to thrive. Long-term sequelae of severe airway obstruction can lead to cor pulmonale and cardiorespiratory arrest.
Interventions aimed at relieving such symptoms vary depending on the severity of the symptoms. Conservative measures for less severe symptoms include prone positioning and use of a nasopharyngeal airway.2 In the event that such conservative treatments are unsuccessful or for more severe symptoms, options for surgical intervention include tongue-lip adhesion, tracheotomy, and mandibular distraction osteogenesis (MDO). Tongue-lip adhesion has been shown to significantly improve obstructive sleep apnea due to micrognathia3 but has generally been less effective than MDO at normalizing obstructive symptoms, particularly in severely symptomatic patients. In addition, tongue-lip adhesion can lead to dysphagia and feeding difficulties.4 Tracheotomy offers a definitive treatment for upper airway obstruction but has associated risks of accidental decannulation or mucous plugging. There is also potential long-term morbidity related to peristomal scarring and tracheal erosion in addition to the need for long-term maintenance and home care.5,6 In recent years, there have been an increasing number of reports on the results of MDO as an alternative to tracheotomy.7- 15
The goals of this study were to assess the surgical success and complication rates of MDO for treatment of severe micrognathia and to identify potential predictors of surgical success and complications. Surgical success was defined as either (1) avoidance of tracheotomy or (2) decannulation among those patients treated initially with a tracheotomy.
This was a retrospective cohort study of all patients who underwent MDO from September 1, 1995, to December 31, 2009, at Cincinnati Children’s Hospital Medical Center. Inclusion criteria included any patients who underwent initial MDO during the study period. All patients were seen through a multidisciplinary craniofacial clinic, and those with a concern for syndromic Pierre Robin sequence were routinely evaluated by clinicians from both the genetics and ophthalmology departments in addition to the craniofacial surgery and otolaryngology departments to ensure proper diagnosis and management. Electronic and paper chart medical records were reviewed for relevant data. Patients who were lost to follow-up after MDO or had incomplete medical records were excluded. Mandibular distraction osteogenesis was performed with short sagittal split osteotomies using primarily external distraction devices, although internal distraction devices were used in a minority of patients based on surgeon preference. This study was approved by the institutional review board of Cincinnati Children’s Hospital.
Variables included as potential predictors of outcome included demographics (sex, age at time of distraction), follow-up time, syndrome presence (categorized as isolated Pierre Robin sequence, craniofacial microsomia [CFM] or Goldenhar syndrome, Treacher-Collins syndrome, and other syndromes), type of initial surgical intervention (tracheotomy vs MDO), length of mandible distracted, number of distractions, and number of subsequent airway procedures (eg, laryngotracheoplasty, endoscopic airway procedures, base of tongue procedures, choanal atresia repair).
While the primary goal of MDO was to improve the retrolingual airway and relieve airway obstruction, the definition of surgical success necessarily differed for patients who were initially treated with tracheotomy prior to MDO compared with those who underwent MDO as an initial procedure. Thus, surgical success was defined as (1) avoidance of tracheotomy among patients who were treated first with MDO and (2) successful decannulation among patients who initially underwent tracheotomy prior to MDO. Complications assessed included open bite deformity, premature bone consolidation, temporomandibular joint (TMJ) ankylosis, facial nerve injury, emergent reintubation, and prolonged intubation. For the purposes of this analysis, the need for repeated distraction more than 30 days after the initial distraction was not considered a complication because it was felt that this need reflected a lack of innate growth of the distracted mandible over time rather than a failure of the initial distraction.
Descriptive statistics are reported as means (SDs) and medians with interquartile ranges (IQRs) or frequencies with percentages. The characteristics of the 2 subgroups defined by initial surgical treatment were compared using the Wilcoxon rank sum test for continuous variables and the χ2 or Fisher exact test for categorical variables. Logistic regression analysis was used to assess the relationship between the potential predictors and each dichotomous outcome of interest (surgical success and occurrence of complications). The analysis of surgical success was stratified by initial surgical intervention (tracheotomy vs MDO), since the definition of outcome differed between these subgroups. For potential predictors of a complication, all complications were grouped together as a dichotomous outcome (any complication or no complication). The entire cohort was included in this regression analysis since the definition of a complication was the same regardless of the initial intervention. For each outcome (surgical success and occurrence of a complication), logistic regression models were constructed to identify potentially important associations between the predictor variables and each outcome of interest. A conservative criterion P = .20 was used as a cutoff for inclusion in subsequent multivariable regression models. An α = .05 was considered for statistical significance in all final models. SAS statistical software (version 9.3; SAS Institute) was used to conduct all analyses.
There were 132 patients who underwent MDO during the study period. Of these, 8 patients were lost to follow-up, and 1 died shortly after distraction owing to congenital heart disease. These patients were excluded from subsequent analysis, leaving 123 patients in the cohort. Patient characteristics for the entire cohort and stratified by initial treatment group are described in Table 1. A slight majority of patients were male, and 56.0% were diagnosed as having an associated syndrome. Treacher-Collins (9.8%) and CFM-Goldenhar (8.9%) syndromes were the most commonly encountered syndromes. Median age at time of distraction was 21 months (range, 6 days–24 years). Sixty-two patients (50.4%) underwent an initial tracheotomy prior to MDO while 61 (49.6%) underwent MDO first. Median follow-up time was approximately 5 years (range, 30 days–16.2 years). The median distraction amount was 22 mm (range, 7-52 mm). One hundred seven patients (87.0%) underwent a single distraction, while the remainder of the cohort required repeated distractions. Most these patients underwent repeated distractions at least 6 months after the initial distraction. Seven patients (5.7%) required repeated distractions within 30 days of the initial procedure. Of these, 4 could be attributed to either hardware malfunction or premature consolidation. Sixty-seven patients (55.0%) did not require any further airway procedures after distraction, although a substantial minority of patients required at least 1 other procedure.
Patients who underwent tracheotomy first were significantly more likely to have an associated syndromic diagnosis compared with those who underwent MDO first (66.0% vs 43.0%, respectively). Patients treated with MDO first were also significantly younger at the time of distraction than those treated with tracheotomy first (median age, 5 months vs 30 months) and required fewer subsequent airway procedures.
Rates of surgical success and complications are described in Table 1. The overall success rate for MDO in the cohort was 75.6%. There was a significant difference in the success rate between patients who underwent tracheotomy prior to MDO (67.7% successfully decannulated) compared with those who underwent MDO first (83.6% avoided tracheotomy; P < .001). In the entire cohort, there were 72 patients who underwent tracheotomy, 62 who underwent tracheotomy as an initial procedure, and 10 who underwent tracheotomy after MDO. Approximately one-third of these were performed at outside institutions prior to referral. Five of the 10 patients who required a tracheotomy after MDO were eventually decannulated. The overall complication rate was 26.8%, with a significantly higher complication rate in the tracheotomy-first subgroup compared with the MDO-first subgroup (38.7% vs 14.8%, respectively; P = .003). In the overall cohort, premature bony consolidation (11.4%), open bite deformity (7.3%), and TMJ ankylosis (4.1%) were the most common complications. Patients who underwent a tracheotomy first had greater rates of premature consolidation (19.4% vs 3.3%; P = .005) and TMJ ankylosis (8.1% vs 0%; P = .06) compared with those who underwent MDO first. Among the 5 patients who developed TMJ ankylosis, 1 was a patient with amniotic band syndrome and bilateral Tessier 7 clefts who required 4 distraction procedures and developed ankylosis after the last distraction. One patient with Catel-Manzke syndrome had preexisting TMJ ankylosis that was thought to be related to her underlying syndrome, and another had Goldenhar syndrome (Pruzansky grade 1). Both of these patients required 3 distractions each. One patient with isolated micrognathia had premature consolidation requiring a second distraction procedure that was complicated by a pin site infection, and the last patient had isolated micrognathia requiring only 1 distraction. In this case, the cause of the TMJ ankylosis was unclear.
In the tracheotomy-first subgroup, univariable logistic regression modeling identified sex, syndrome diagnosis, and age at distraction as potentially important predictors of surgical success (Table 2). When adjusting for sex and age at distraction, patients with CFM–Goldenhar syndrome had the lowest probability of surgical success (OR, 0.07 [95% CI, 0.009-0.52]) compared with patients with isolated Pierre Robin sequence (Table 3). To better illustrate the impact of these different variables on the probability of surgical success, Figure 1 shows the modeled probabilities of success as a function of age and syndrome diagnosis, stratified by sex. For both male and female patients, at any given age, the probability of surgical success is significantly worse for patients with CFM–Goldenhar syndrome compared with any other syndromic diagnosis. In contrast, patients with isolated Pierre Robin sequence have the greatest probability of success. For example, a 10-year-old girl with Goldenhar syndrome who went through a tracheotomy before MDO has a 60% probability of surgical success. For all patients, the probability of success seems to increase with greater age at the time of distraction.
In the MDO-first subgroup, the only variable associated with surgical success in univariable regression analysis was number of other airway surgical procedures (Table 4). Thus, multivariable regression analysis was not performed in this subgroup. In the univariable regression model, patients who had undergone fewer than 2 airway procedures had 7 times greater odds of success compared with those requiring more than 2 procedures. Among the 10 patients who required a tracheotomy after an initial MDO, airway pathology contributing to failure of initial MDO in these patients included persistent glossoptosis or lingual tonsil hypertrophy (8 patients), tracheal stenosis (2), and choanal atresia (1). Five of 10 were subsequently successfully decannulated, but all required additional procedures to achieve decannulation, including lingual tonsillectomy and/or base of tongue reduction (3 patients), endoscopic suprastomal granulation tissue removal (2), LeFort I bimaxillary advancement (3), or choanal atresia repair (1).
When we examined potential predictors of a complication, univariable regression analysis demonstrated an association between occurrence of a complication and patients who underwent a tracheotomy prior to MDO (OR, 2.9 [95% CI, 1.2-7.1]), increasing length of follow-up (OR, 1.2 [95% CI, 1.0-1.3]), and patients who required 2 or more airway procedures (OR, 3.4 [95% CI, 1.4-8.4]) compared with patients with 0 or 1 other procedure. In a multivariable regression model adjusting for all these variables, only length of follow-up (OR, 1.2 [95% CI, 1.0-1.3]) and number of other airway procedures (OR, 3.2 [95% CI, 1.2-8.6]) remained significantly associated with occurrence of a complication. Figure 2 demonstrates the modeled probabilities of a complication with MDO in the 2 initial treatment subgroups as a function of these 2 independent predictors. One can see that in both the MDO-first subgroup and the tracheotomy-first subgroup, the probability of a complication increases with length of follow-up time. In addition, for any given follow-up duration, there is a greater probability of a complication when 2 or more other airway procedures are required.
The application of distraction osteogenesis to the mandible for the treatment of symptomatic micrognathia was first described by McCarthy et al16 in 1992, and since that time there have been increasing reports of the outcomes of this intervention in the pediatric population. Owing to the heterogeneous nature of the patient population that experiences symptomatic micrognathia severe enough to warrant surgical intervention, definitions of success have varied depending on the specific study. Given this context, success rates for MDO in improving or relieving airway obstruction due to micrognathia have been reported to range from 63% to 100% depending on the definition of success used.8- 11,17- 20 In a recent meta-analysis, Ow and Cheung19 found a 91% rate of prevention of tracheotomy among neonates undergoing MDO as an initial procedure and a 78% rate of decannulation among patients with existing tracheostomies who underwent MDO.
This study reports the outcomes of MDO in one of the largest cohorts of patients treated for symptomatic micrognathia. The relatively large cohort provided greater statistical power than most previous reports and facilitated the investigation of potential predictors of surgical success and complications. Overall, we found a high rate of surgical success among patients undergoing MDO, especially among those undergoing MDO as an initial procedure to treat symptomatic micrognathia. The 84% rate of avoidance of tracheotomy is slightly lower than the success rates in recent reports focusing on the outcomes of neonatal MDO.8,9,17,18 However, this may in part reflect the fact that there was a greater proportion of syndromic diagnoses (56% overall, 43% among patients treated with MDO first) than in other recent studies in which the proportion of syndromic diagnoses ranged from 24% to 35%.8- 11,20 The most frequently encountered syndromic diagnosis among all patients with micrognathia is Stickler syndrome.2,21,22 However, in our study the 2 most frequent syndromic diagnoses were Treacher-Collins and CFM–Goldenhar syndromes. It is unclear why we observed such a high prevalence of these syndromes relative to Stickler syndrome, but this may represent a relatively skewed population of patients who are referred from outside the local area for tertiary and quaternary care. These syndromes often demonstrate more severe micrognathia than in isolated Pierre Robin sequence or even other syndromic forms of micrognathia23 and are therefore more likely to have poorer outcomes.18 In our cohort, 4 of the 11 patients with CFM–Goldenhar syndrome and 2 of the 12 patients with Treacher-Collins syndrome had Pruzansky grade 3 classification with an absent mandibular condyle, and it is possible that this may have also limited the effectiveness of MDO in these patients. Given the greater percentage of these syndromes in our cohort, this may in part explain why the overall rate of successful MDO was slightly lower than in other published reports.
When we examined the potential predictors of surgical success, the different definitions of success for the 2 subgroups (decannulation vs avoidance of tracheotomy) necessitated separate regression analyses based on initial intervention (tracheotomy first vs MDO first). Thus, we cannot draw conclusions regarding predictors of success across both subgroups. However, we can reasonably conclude that among patients who required a tracheotomy as an initial procedure, patients with CFM–Goldenhar syndrome seem to have a far worse chance of success with subsequent MDO than patients with isolated Pierre Robin sequence. This association does not seem to hold among patients who underwent MDO first in the absence of a tracheotomy.
Compared with patients treated with MDO first, those treated with tracheotomy first had significantly more syndromic diagnoses and were older at the time of MDO. Many of these patients had additional comorbidities, such as multilevel airway obstruction, neuromuscular compromise, and other medical comorbidities that necessitated a tracheotomy as a definitive treatment for relief of their airway obstruction. Achieving decannulation in this subgroup necessitated significantly more adjunct airway procedures, such as laryngotracheoplasty or endoscopic airway procedures, compared with the subgroup that underwent MDO first. The finding in univariable regression analysis that patients who underwent tracheotomy first had greater odds of a surgical complication makes intuitive sense in this context. The fact that this association was no longer significant in a multivariable model that also adjusted for the number of adjunct airway procedures suggests that the association between initial tracheotomy and complication is perhaps mediated by the need for additional procedures in order to achieve decannulation. Taken together, these findings are consistent with our hypothesis that patients selected for tracheotomy rather than MDO as an initial procedure were more medically complex and required more airway interventions in order to achieve a stable airway. This form of selection bias may have contributed to their worse rates of success and complication and also creates an inherent confounding by indication when making comparisons between these 2 subgroups (MDO first vs tracheotomy first).
Limitations to our study include the retrospective nature of the data collection, lost or missing data, and the inherent confounding by indication that occurs in comparisons between the MDO-first subgroup vs the tracheotomy-first subgroup. In addition, prior to 2002, when neonatal MDO became part of routine practice at our institution, there may have been an even stronger selection bias toward tracheotomy as an initial procedure. To address this possibility, a secondary analysis was performed with the data set restricted to only those patients who were treated with MDO after 2002, and the results were essentially unchanged, suggesting minimal impact of this possible bias. Regardless, the issue of confounding by indication is an important one, which is why the outcomes of surgical success were addressed in separate analyses in this study. Within the context of this broad retrospective study, we did not attempt to directly compare the 2 primary interventions of tracheotomy and MDO except with respect to overall rates of complications. However, it is clear from our analysis that patients who were syndromic or had additional airway disease abnormalities were more likely to have an initial tracheotomy to address multilevel disease. Mandibular distraction osteogenesis should be considered only as an initial procedure in a setting in which there is no other indication for tracheotomy, such as multilevel airway obstruction, ventilator dependence, or persistent dysphagia and aspiration. Because patients with CFM-Goldenhar syndrome were found to have independently lower odds of success with MDO than other syndromic conditions, we would not recommend primary MDO in these patients who have clinically significant respiratory distress due to micrognathia. A proposed treatment algorithm is presented in Figure 3.
Our study evaluated a large cohort of children that had undergone MDO. In our study we demonstrated a high rate of surgical success for MDO with a low rate of complications, particularly among patients treated with distraction as an initial procedure without an existing tracheotomy. However, it is clear that patients treated with tracheotomy initially are more likely to be syndromic and complex, requiring greater numbers of distractions and airway procedures. In addition, patients with CFM–Goldenhar syndrome have a decreased probability of surgical success compared with patients with other types of syndromes or nonsyndromic Pierre Robin sequence.
Submitted for Publication: September 6, 2013; final revision received December 14, 2013; accepted January 6, 2014.
Corresponding Author: Derek J. Lam, MD, MPH, Department of Otolaryngology–Head and Neck Surgery, Oregon Health and Science University, 3181 SW Sam Jackson Park Rd, PV-01, Portland, OR 97239-3098 (email@example.com).
Published Online: February 27, 2014. doi:10.1001/jamaoto.2014.16.
Author Contributions: Dr Lam had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Lam, Shikary, Uribe-Rivera, de Alarcon, Gordon.
Acquisition of data: Lam, Shikary, Uribe-Rivera.
Analysis and interpretation of data: Lam, Tabangin, Shikary, Meinzen-Derr, Billmire.
Drafting of the manuscript: Lam, Meinzen-Derr.
Critical revision of the manuscript for important intellectual content: Lam, Tabangin, Shikary, Uribe-Rivera, de Alarcon, Billmire, Gordon.
Statistical analysis: Lam, Tabangin, Shikary, Meinzen-Derr.
Administrative, technical, and material support: Uribe-Rivera, Billmire.
Study supervision: Lam, de Alarcon, Billmire, Gordon.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented at the American Society of Pediatric Otolaryngology Spring Meeting; April 19-22, 2012; San Diego, California.