Late-Onset Laryngomalacia: A Variant of Disease

Objective  To identify and describe the features of laryngomalacia (LM) in a cohort of older children, with the goal of providing an approach to diagnosis and management of these patients. Laryngomalacia is a common congenital disorder characterized by decreased laryngeal tone, supraglottic collapse, and stridor during inspiration and is rarely seen in older children. However, the presence of LM in this population may be obscured by related but uncommon clinical features.

Design  Prospective collection and retrospective evaluation of older children with evidence of LM from 1998 to 2005.

Setting  Two tertiary pediatric institutions.

Patients  Prospective data collection of 239 patients with LM, including 222 with congenital LM and 17 diagnosed as having LM when they were older than 2 years and without a medical history of prior disease or symptoms—late-onset LM (mean age at onset, 6.6 years).

Intervention  Supraglottoplasty.

Main Outcome Measure  Symptom improvement.

Results  Patients with late-onset LM were classified into 3 categories according to their symptom complex; those presenting with feeding-disordered LM (n = 7; mean age at onset, 3.3 years), sleep-disordered LM (n = 7; mean age at onset, 6.3 years), and exercise-induced LM (n = 3; mean age at onset, 15 years). Stridor was rarely present except in patients with exercise-induced LM during strenuous activity. Profound arytenoid redundancy and prolapse was discovered in all patients during nasolaryngeal endoscopy. Typical anatomic features of congenital LM (shortened aryepiglottic folds or retroflexed epiglottis) were not discovered. No patient had a history of neuromuscular disease. Supra-arytenoid reduction (supraglottoplasty) led to clinical cure in all patients (mean duration of follow-up, 7.4 months). Prior adenotonsillectomies were performed in 5 patients with sleep-disordered LM. This did not improve symptoms, and these patients displayed evidence of LM as the source of obstruction.

Conclusions  Physicians should consider late-onset LM as a potential cause of feeding difficulties in toddlers, sleep apnea in children, and exercise intolerance in teenagers. As in infants with LM, supraglottoplasty improves late-onset disease.

The term laryngomalacia (LM) was coined in 1942¹ to describe reduced laryngeal tone and inward prolapse of supraglottic structures in infants with congenital inspiratory stridor. Congenital LM (CLM) is now considered to be the most common cause of stridor and laryngeal obstruction in the newborn, with clinical ties to gastroesophageal reflux disease (GERD) and laryngopharyngeal reflux (LPR).^2,3 The severity of CLM can be broad and unpredictable, with intermittent cyanosis, retractions, coughing, choking, and reflux accompanying moderate to severe disease. Fortunately, most cases of LM are self-limited by 1 to 2 years of age with expectant management of feeding modification and reflux therapy.⁴ Supraglottoplasty is employed when indicators of severe upper airway obstruction and failure to thrive appear.⁵

Laryngomalacia has become a ubiquitous term to describe a floppy-appearing larynx despite the age of the patient. However, when identified in older patients, the symptom complex is often different from that of patients with congenital disease. For example, LM has been described in adolescent and adults patients with “acquired” and “exercise-induced” LM (EILM).^6-12 Similarly, “state-dependent” LM has been reported in infants and toddlers presenting with sleep-induced stridor, upper airway obstruction, and apnea.^13,14 Although the larynx in older children can behave similarly to that of infants with CLM, the same symptoms are often absent. Thus, indications for medical and surgical treatment remain unclear in this patient population. Our objective was to characterize the clinical features of older children with LM discovered among a large cohort of children with this diagnosis. Specific attention was paid to medical and surgical interventions to help elucidate an ideal approach to these children.

With institutional review board approval, data were collected at 2 tertiary care pediatric referral centers from 1998 to 2005 for 239 patients diagnosed as having LM. Children included in this study were those with an LM-appearing larynx diagnosed at an age older than 2 years. Patients were excluded if they had neuromuscular disease or had symptom onset before age 2 years. To determine how these children with late-onset LM differed from the typical patient with CLM, the database was reviewed for clinical symptoms and demographic, diagnostic, and therapeutic variables. Data collected on patients with CLM have previously been published.¹⁵ All children had a diagnosis established by flexible awake laryngoscopy and/or intraoperative laryngoscopy during light sedation. Those with EILM were also evaluated by flexible laryngoscopy during exercise while achieving 85% aerobic capacity, or during a hyperventilative state for 10 minutes. The database was reviewed for clinical and diagnostic studies of GERD and LPR. Medical and surgical treatment strategies, along with their outcomes, were retrospectively reviewed.

Seventeen patients with a mean age of 6.6 years were diagnosed as having late-onset LM based on laryngoscopy findings of an LM-appearing larynx. After review of each child's presenting symptom, the patients were classified into 3 groups by presenting symptom: (1) feeding-disordered LM (FDLM), (2) sleep-disordered LM (SDLM), and (3) EILM. No patient had a prior diagnosis or symptom consistent with CLM or neuromuscular disease or presented with the stridor typical of CLM. The findings for each patient group and treatment strategy follow (Table).

Seven patients with dysphagia and a feeding disorder were diagnosed as having an LM-appearing larynx. The mean age of the group was 3.3 years. Their presenting symptoms included persistent coughing and choking during feeding, weight loss, and failure to thrive. The mean percentile for weight for these patients was the fifth percentile. These patients were referred for flexible endoscopic evaluation of swallowing (FEES) to evaluate for laryngeal anomalies or problems with airway protection in the context of a feeding problem. The FEES examination findings included an LM-appearing larynx with redundant posterior glottic tissue and supra-arytenoid tissue prolapse into the glottic inlet obscuring full visualization of the vocal folds. This finding was seen at rest but was most pronounced during the act of swallowing. A feeding assessment with a thin liquid demonstrated laryngeal penetration in all patients, whereas aspiration beyond the vocal folds was seen in 3. All children demonstrated increased work of breathing while swallowing, with 4 children exhibiting mild inspiratory stridor. In contrast to CLM, patients with FDLM did not demonstrate inspiratory stridor at rest. One patient underwent adenotonsillectomy for signs and symptoms of obstructive sleep apnea (OSA) and dysphagia prior to establishing the diagnosis of FDLM. The obstructive symptoms resolved, but the feeding symptoms persisted (duration of follow-up, 4 months).

Both GERD and LPR symptoms were a common denominator in these patients. Diagnostic evidence of GERD was confirmed in all 7 patients by 24-hour pH-metry. Patients were initially treated with antireflux medical therapy using a daily proton pump inhibitor and nightly histamine 2 receptor antagonist with behavioral precautions for a median duration of 5 months. Recalcitrant signs and symptoms, poor weight gain, and endoscopic evidence of persistent arytenoid prolapse led to supraglottoplasty in each patient. Intraoperative findings of redundant supra-arytenoid tissue during microlaryngoscopy and rigid bronchoscopy were consistent with findings from preoperative laryngeal examinations. Reduction of posterior and superior arytenoid tissue was performed using standard supraglottoplasty techniques and microlaryngeal instruments. Postoperative FEES examination showed resolution of feeding disturbances and improved endoscopic findings. All patients achieved resolution of feeding symptoms and gained weight. At the time of the last visit, the mean percentile for weight was the 40th percentile (duration of mean follow-up, 10.4 months).

Seven patients were diagnosed as having late-onset SDLM. The mean age of the group was 6.3 years. All children presented with parental reports of sleep-disordered breathing and symptoms consistent with OSA that included respiratory pauses, gasping, restless sleep, and suprasternal and subcostal retractions. Five of these patients had adenotonsillectomy surgery prior to the diagnosis of SDLM. Two parents reported stridor during sleep-dependent upper airway obstruction. Two patients were described as having mild complaints of EILM with shortness of breath during strenuous activity. All children were small for their age and had a mean weight in the 37th percentile. Three children were diagnosed as having clinically significant GERD by 24-hour pH-metry. All 6 were treated for GERD with a combination of daily proton pump inhibitor and nightly histamine 2 receptor antagonist therapy for a mean of 3 months. All children had persistent symptoms despite medical treatment of their GERD. Preoperative polysomnography confirmed the diagnosis of sleep apnea with intermittent hypoxia (mean lowest arterial oxygen saturation, 86%), hypercapnia (mean carbon dioxide level, 58 mm Hg), and intermittent hypopnea and apnea (mean apnea-hypopnea index, 6). Preoperative awake nasolaryngoscopy revealed a normal-appearing larynx. To determine the level of airway obstruction, the patients were examined with flexible nasolaryngoscopy during stage 2 of anesthesia and by rigid microlaryngoscopy and bronchoscopy during stage 3 while maintaining spontaneous breathing. These examinations demonstrated arytenoid tissue redundancy and prolapse (Figure) during both stages of anesthesia. Features of CLM, such as shortened aryepiglottic folds and retroflexed epiglottis, were not identified. Other levels of airway obstruction were not seen in the 5 patients who had undergone previous adenotonsillectomy surgery. One patient demonstrated mild tongue base collapse and adenoid hypertrophy and underwent adenoidectomy and tongue base resection at the same anesthetic setting. While anesthetized, the patients exhibited airway obstruction, intermittent stridor, and retractions. Owing to persistent symptoms and the fact that OSA could not be attributed to sources beyond the supraglottic larynx (arytenoid processes), all patients underwent superior arytenoid tissue reduction. Each patient has remained symptom free following supraglottoplasty (mean duration of follow up, 7.4 months) and has required no further intervention for their OSA. Postprocedure polysomnography data are not available because all patients dramatically improved clinically. The mean weight percentile for the group had increased to the 55th percentile at the time of last follow-up.

We suspected EILM in 3 patients aged 14, 15, and 16 years. Each patient presented with a history of EI symptoms of inspiratory stridor, shortness of breath, and retractions with strenuous activity. Symptoms were not present during periods of rest or sleep. Consistent with EILM and not EI asthma, all patients failed to respond to inhaled β-agonist or cromolyn sodium therapy. Two patients had symptoms and diagnostic evidence consistent with GERD based on 24-hour pH-metry. All underwent medical treatment of GERD using daily proton pump inhibitor and nightly histamine 2 receptor antagonist. The diagnosis of EILM was confirmed in 2 patients by exercise treadmill testing during simultaneous flexible laryngoscopy while exercising to 85% of aerobic capacity for 10 minutes. A hyperventilation study was performed in 1 patient by asking the patient to perform deep and rapid respirations to simulate EI tachypnea and hyperpnea while undergoing flexible laryngoscopy for 10 minutes. Arytenoid rotation and supra-arytenoid prolapse into the laryngeal inlet was identified in all 3 patients during these studies. In addition, all patients demonstrated normal vocal fold mobility, oropharyngeal and hypopharyngeal anatomy, and normal laryngeal dynamics during rest. Supraglottoplasty, with removal of the supra-arytenoid tissue only, was performed in all patients and resulted in resolution of symptoms during strenuous activity and exercise (mean duration of follow-up, 7 months). Postoperative exercise laryngoscopy was not performed because all patients' symptoms resolved.

In this study, we describe a subset of older children with sporadic LM. Three distinct patient populations were identified based on presenting symptoms, because patients of similar age coalesced within each group. Supraglottic collapse was manifested as FDLM in toddlers, SDLM in school-aged children, and EILM in teenagers. Symptom overlap existed in 2 patients with SDLM who had mild but inconclusive symptoms of EILM. A diagnosis of GERD was confirmed in 100%, 43%, and 67% of patients with FDLM, SDLM, and EILM, respectively (Table). Regardless of age and presenting symptoms, functional laryngeal studies illustrated obstruction at the arytenoid and supra-arytenoid levels with a normal-appearing epiglottis and aryepiglottic fold. This is in contrast to epiglottic and aryepiglottic collapse often seen in CLM but is consistent with previous descriptions of older children and adults with variants of LM.^6,8,13,16,17 Conservative medical management of GERD and LPR in this patient population was unsuccessful. However, trimming of the supra-arytenoid tissue relieved patient symptoms in each late-onset LM group and help confirm LM as the source of disease. This also illustrated that late-onset disease is unlikely to resolve over time or respond to GERD treatment as commonly seen in severe CLM.^5,18 Although presenting symptoms differed among groups with late-onset LM, these patients responded similarly to therapeutic intervention.

This is a large series of patients diagnosed as having LM at an older age and evidence of 3 disparate disease presentations. We offer the term late-onset as a way of delineating these various presentations of LM from congenital disease that is found in the infant. This does not necessarily imply late development of a congenital laryngeal disruption but rather a wholly different entity that is likely underrecognized and more common than what is currently reported. In contrast to CLM, the site of disease was invariably discovered at the arytenoid and supra-arytenoid level in these older children who had no previous symptoms or diagnosis of LM (Table). Decreased laryngeal tone and supraglottic prolapse may thereby be reasonably considered a source of feeding difficulties, sleep apnea, and EILM in older children. Because laryngeal dynamics and symptoms improve with supraglottoplasty in both patients with congenital disease and in those with late-onset disease (FDLM, SDLM, EILM), we find comfort in describing these entities as variants of LM. This may aid in the diagnosis and treatment of atypical presentations of LM and potentially illustrate disease mechanisms of LM in the future.

It remains unclear why some patients develop LM, or a flaccid larynx, after infancy. Previously reported cases of late-developing, or acquired, LM have been associated with neurologic injury and resultant neuromuscular hypotonia.^12,19 However, no child in this study had evidence of such neurologic deficits.

In CLM, feeding difficulties are the second most common complaint next to stridor.⁴ Similarly, feeding issues were the most frequent presenting symptom in our series of patients. This was confined to patients of toddler age. Symptoms were consistent with those found in CLM and included choking and coughing during feedings, regurgitation, weight loss, and difficulty gaining weight. In contrast, no individual had current or prior evidence of stridor, a symptom found in nearly 100% of patients with CLM. To our knowledge, this is the first report of feeding difficulties attributed to LM in older children without a prior diagnosis of LM at infancy. This may be a more frequent source of feeding issues in toddlers than is currently charged. The discovery of arytenoid prolapse on FEES and the resultant impact of supraglottoplasty on disease resolution provide credence to LM as the appropriate diagnosis in these patients. Arytenoid prolapse classically occurs during passive respiration, is exacerbated during feeding, and results in stridor in patients diagnosed as having CLM. In contrast. this sign was evident only during the feeding phase of the examination of these children with late-onset disease. Recognition of this specific finding is important in identifying patients with LM who present later in life with feeding difficulties.

Both GERD and LPR have been implicated in the clinical spectrum of patients with CLM and symptoms of regurgitation, coughing and choking with feeding, and weight loss.^2,3,15,20 In this study, GERD was diagnosed in each patient with late-onset LM and with feeding difficulties. It may, therefore, be reasonable to surmise that LPR is a contributing factor to their disease. Also, GERD has been shown to stimulate edema of posterior glottic and supra-arytenoid tissue.²⁰ This is thought to contribute substantially to prolapsing redundant structures during forceful or discoordinated inspiration and swallowing in CLM. However, in the older child, laryngeal dynamics and size are improved relative to the immature larynx with congenital disease. Because the airway is larger, symptoms of airway obstruction would be less pronounced in this patient population. This likely explains why these children did not have stridor at rest and only mild respiratory symptoms during swallowing. Although sensory testing data were not integrated in this study, we infer that the effects of acid exposure to the posterior glottis may lead to interruption of normal sensation and swallowing function. This has been demonstrated in 2 previous investigations^15,21 of infants and children with GERD-associated feeding disorders. Greater negative inspiratory pressures required to overcome these mild levels of laryngeal obstruction can lead to a cyclical increase in reflux contents transmitted to the posterior and supraglottic larynx and contribute to feeding difficulties. Logically, this process reverses once the level of airway obstruction is eliminated (supraglottoplasty), as seen in our patient series with late-onset LM.

In this study we describe 7 patients with SDLM and evidence of arytenoid obstruction as the source of disease. Five patients underwent adenotonsillectomy for symptoms of OSA; however, persistent symptoms and dynamic prolapse of the supra-arytenoid tissue were found during intra-operative diagnostic laryngoscopy. Surgical correction of the larynx ultimately led to symptom resolution. Interestingly, symptoms and laryngeal findings in our 7 patients with SDLM are consistent with those described in previously published reports of children with LM-associated sleep apnea. In particular, Smith et al¹⁴ discovered arytenoid and supra-arytenoid prolapse as the source of obstruction during spontaneous breathing under anesthesia in each of 4 otherwise healthy patients presenting with sleep-associated stridor. Similarly, Amin and Isaacson¹³ described the inward prolapse of arytenoid structures found during anesthesia in 5 infants who had stridor while asleep but a normal-appearing larynx when examined awake. The mean age of their patient group was 6 months, whereas our patients with SDLM were school-aged children. Regardless of age, both studies demonstrate the value of endoscopy under anesthesia to identify laryngeal obstruction when it is difficult to locate during the awake state.

The cause of pediatric OSA can be related to either fixed (eg, adenotonsillar hypertrophy) or dynamic (eg, LM) upper-airway abnormalities. Naturally, dynamic problems more commonly occur in patients with systemic hypotonia. Goldberg et al²² illustrated 7 of 17 patients with sleep-onset LM who had underlying hypotonic disorders. Although fixed abnormalities (adenoid hypertrophy) were found in combination with LM in several patients in their study, they also identified 10 anatomically normal individuals with LM-associated OSA. This patient population can be equated with those in our series because LM-associated OSA occurs following removal of tonsil and adenoid disease. Why these children have a normal-appearing laryngeal examination when awake but LM when asleep is unclear.

Reduced tone of the pharyngeal dilators and neuromuscular support system of the supraglottic larynx (stylopharyngeus, palatopharyngeus, hypoglossus, and digastric muscles) likely contributes to the inspiratory collapse.²³ This is found in older children with neuromuscular hypotonia where underlying muscle and central control to maintain airway patency is reduced.^22,24,25 Although central reductions in tone occur during sleep, this does not explain LM-associated OSA in older children when neuromuscular disorders are absent, as seen in our patients. In a child without an underlying neurologic problem, one would expect to see normal laryngeal tone when the patient is examined while awake. It is possible that the children with SDLM experience nighttime GERD that is exacerbated during sleep, especially under conditions of increased respiratory effort with mild upper airway obstruction. This results in an increase in supraglottic edema, which may explain some of the clinical examination findings of redundant supra-arytenoid tissue and airway obstruction seen during a sleeplike state. This is an unlikely but plausible explanation for the atypical onset of LM-associated OSA. We surmise that these patients were predisposed to mild reductions in laryngeal tone that led to a worsening and cyclical state of supraglottic edema, GERD, and laryngeal tone. As described in the “Feeding-Disordered LM” subsection, removing redundant supra-arytenoid tissue, while maintaining control of GERD, can lead to improved supraglottic space, dynamics, sensitivity, and subsequent tone of the larynx during sleep. The result is symptom resolution, as seen in our series of patients.

Another potential mechanism may be related to the impact of hypoxia and hypercarbia on laryngeal function. Hypoxia can exert its effects by regulating respiration and laryngeal tone. Hypoventilation, hypercapnia, and hypoxia have been proven to lead to decreased laryngeal tone and neuromuscular transmission between the supraglottic larynx and centers for respiration in the brainstem.^26,27 The patients in this study showed periods of intermittent hypoxia and hypercapnia demonstrated by polysomnography, thereby providing another plausible explanation for the finding of decreased laryngeal tone and LM only during sleep.

Smith et al¹⁰ first reported EILM in 1995. Since then, several reports of EILM have been published worldwide.^{7,8,10,16,17,28,29} Similar to our findings, supra-arytenoid tissue has been reported as a primary site of obstruction in EILM, although the aryepiglottic folds may also be involved. Exercise-induced LM is a disease of adolescence and most frequently affects athletic girls.^16,28 Stridor accompanied by respiratory retractions during strenuous exercise is characteristic, but patients may be misdiagnosed with either asthma or EI asthma. Failure to improve symptomatically with β-agonist or cromolyn therapy suggests the presence of EILM. These patients will also display immediate improvement in obstructive symptoms as exertion is decreased.

Although not performed in this study, pulmonary function tests can help differentiate EI asthma from EILM. Spirometry in patients with EI asthma demonstrates a flattened expiratory curve and a normal inspiratory curve, whereas patients with EILM display the opposite: a flattened inspiratory curve but normal expiratory curve. Simultaneous performance of nasolaryngoscopy and treadmill testing (or hyperventilation studies) will confirm disease and assist in eliminating other possible etiologies (paradoxical vocal cord motion).

Herein, we describe 3 patients with EILM with evidence of stridor, dyspnea, and arytenoid prolapse during strenuous activity. Treadmill and hyperventilation studies confirmed disease in each patient. As suggested by Abu-Hasan et al,²⁸ we suspect the identification of more patients with EILM will occur as awareness increases and the use of pulmonary function tests and flexible nasolaryngoscopy during treadmill testing are employed during the evaluation of patients with suspected EI asthma.

There is limited information available on laryngeal activity during exercise. It is evident, however, that laryngeal dynamics are altered during exercise. The glottic area has been shown to increase during both inspiration and expiration in the healthy individual during strenuous activity.^30,31 As the exercise load increases, so does glottic dilatation, anterior laryngeal rotation, and epiglottic flattening against the base of the tongue.³¹ In contrast, Beaty et al³¹ illustrated that in patients with EILM, an increase in pulmonary load results in exaggerated resistance at the supraglottic level and leads to inward prolapse of supraglottic structures. Glottic diameter subsequently decreases as the arytenoids, unable to maintain their position, rotate medially into the laryngeal lumen. Anatomic dimensions (taller arytenoid and cuneiform cartilages), supraglottic edema, or altered laryngeal muscular tone may contribute to susceptible larynges in patients with EILM to this process.³¹ Furthermore, as a result of the Bernoulli effect, greater air velocity and negative pressure at the laryngeal inlet can lead to worsening symptoms in these susceptible patients.

It is well documented that strenuous exercise induces GERD. This may be explained by a reduction in gastric blood flow of up to 80%, leading to reduced gastric emptying and reduced motor function and pressures of the esophagus during exercise. Reflux is exacerbated in the upright position. In addition, strained body movements during exercise may cause elevations or reversal of gastric and esophageal fluid to the laryngeal inlet.³² This may be increased with body distortions during exercise. Two of the 3 patients in our series demonstrated reflux by pH studies. It is thereby possible that in susceptible adolescents this contributes to LM during exercise. Again, the cyclical nature of GERD and LPR is increasing laryngeal edema, leading to reductions in laryngeal diameter and increases in turbulent airflow and further irritation of laryngeal mucosa. Until the patient is at rest, the diameter of the airway continues to diminish. Surgical correction may improve outcomes by extracting a component of the cycle. To date, all reported cases of EILM have required supraglottoplasty to achieve cure and led to symptom resolution in each patient of this series. Interestingly, EILM exists more commonly in adolescent patients and may also be explained by the alterations in laryngeal anatomy during pubescent growth; however, further investigations are required to better elucidate this possibility.

In conclusion, the presence of LM in the older child may have similar anatomic and clinical features but a distinctly different symptom presentation than CLM. The presence of LM in older children and adults has recently been described in several case reports and series. To our knowledge, this is the first study to identify 3 clearly disparate patient populations with the onset of LM in patients older than 2 years. Late-onset LM was identified as the cause of FDLM in toddlers, SDLM in young children, and EILM in adolescents. Age-specific symptoms are apparent in children with late-onset disease. The site of obstruction at the supra-arytenoid level was consistent among late-onset groups because supra-arytenoid reduction seems to be necessary for symptom resolution in FDLM, SDLM, and EILM.

Correspondence: Gresham T. Richter, MD, Department of Pediatric Otolaryngology, Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, ML 2018, Cincinnati, OH 45229-3039 (gresham.richter@cchmc.org).

Submitted for Publication: May 10, 2007; final revision received September 29, 2007; accepted September 30, 2007.

Author Contributions: Drs Richter, Rutter, deAlarcon, Orvidas, and Thompson had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Richter, Rutter, deAlarcon, Orvidas, and Thompson. Acquisition of data: Richter, Rutter, and Thompson. Analysis and interpretation of data: Richter, Rutter, deAlarcon, Orvidas, and Thompson. Drafting of the manuscript: Richter, Rutter, and Thompson. Critical revision of the manuscript for important intellectual content: Richter, Rutter, deAlarcon, Orvidas, and Thompson. Administrative, technical, and material support: Richter. Study supervision: Richter, Rutter, deAlarcon, Orvidas, and Thompson.

Financial Disclosure: None reported.

Previous Presentation: This article was presented at The American Society of Pediatric Otolaryngology 2007 Annual Meeting; April 28, 2007; San Diego, California.