Acquired Laryngotracheal Stenosis

Management divides broadly into two groups of procedure. Cricotracheal resection excises the stenotic segment wholly and performs an end-to-end anastomosis. Laryngotracheal reconstruction instead expands the laryngeal framework, splitting the cricoid cartilage and reconstructing it with a combination of stents and grafts. Everything else in the operative decision follows from which of those two the child needs. [1]

Resection achieves high rates of decannulation but is an extensive procedure compared with reconstruction, because of the tracheal mobilisation it requires, and a significant length of trachea may need excising if the tracheostomy stoma is included. Reconstruction gives high decannulation rates in grade II and grade III stenosis but tends to be less successful in more severe disease. [1]

Resection is the more successful option in those severe cases, but with one anatomical precondition that governs its use: there must be a clear margin between the stenosis and the vocal cords. Where the stenosis reaches the glottis, resection is not available and expansion is what remains. [1]

The introduction of prolonged endotracheal intubation in the 1960s produced a dramatic rise in paediatric subglottic stenosis. The scale of the improvement since is measurable: the tracheostomy rate in initial studies was around 20% and has fallen over the decades to between 1% and 8%. Excluding very low birth weight infants under 1.5 kg, the incidence is less than 1%. [1]

That fall is likely attributable to using the smallest endotracheal tube size that still permits adequate ventilation. The target is explicit: a tube should ideally leak air around it with subglottic pressures below 20–25 cm of water. The replacement of red rubber endotracheal tubes with polyvinylchloride tubes contributed as well. [1]

Fixation matters as much as the tube. Poor fixation causes movement and increases the chance of stenosis, which is the argument for preferring nasotracheal over orotracheal intubation — less tube movement, more secure fixation. Continuing improvements in airway support, high-flow nasal cannulae and nasopharyngeal continuous positive airway pressure, have significantly reduced the incidence further. [1] [2]

That success changes who now presents. The typical child with subglottic stenosis today is more likely to have genetic abnormalities or to have been born extremely prematurely, which increases the complexity of the population rather than reducing it. Acquired airway stenosis continues to result in high rates of morbidity and mortality despite the falling incidence. [2]

Airway stenosis can occur in the supraglottis, glottis, subglottis or trachea, but it is the subglottis that is most commonly affected in children, because in infants and young children the subglottis is the narrowest part of the airway. In the preterm infant subglottic stenosis is defined as a lumen under 3 mm in diameter, and under 4 mm in a full-term neonate. [1]

A useful bedside guide converts those figures into equipment: the outer diameter of a 3.0 mm endotracheal tube is 4.2 mm. Subglottic stenosis may be acquired or congenital, acquired being far more common and usually secondary to iatrogenic injury from prolonged or repetitive intubation. [1]

Congenital disease is associated with CHARGE syndrome, 22q11 deletion, trisomy 21 and developmental failures during pregnancy. It is the second most common congenital laryngeal anomaly after laryngomalacia, and an elliptical cricoid shape is not an uncommon finding in it — a morphology worth recognising endoscopically because it changes what expansion can achieve. [1]

The original classification proposed by Cotton was revised by Myer and Cotton and is the one most commonly used today. Grade I is lumen obstruction under 50%. Grade II is 51–70% and grade III is 71–99%. Grade IV is complete luminal obstruction. The grades are not descriptive labels but the direct determinant of which operation is offered. [1]

Staging is done by sizing the airway with endotracheal tubes, graded against what is age-appropriate for the individual child rather than against an absolute. The largest tube that can be placed with an air leak at less than 20 cm of water pressure is recorded and staged on that scale. Mild grade I and II stenoses are generally managed non-operatively; grades III and IV typically require surgery. [1] [2]

One limitation of the system must be stated whenever it is used: the Myer–Cotton grading is not designed to address supraglottic, glottic or tracheal stenosis. Grading those levels on this scale is a category error — supraglottic collapse, anterior and posterior glottic stenosis and tracheal stenosis are each separate problems with separate operations. [5]

The primary aim of intervention is to prevent the need for tracheostomy, or to achieve decannulation where a tracheostomy tube is already in situ. This is a highly complex group with highly individualised needs, so a one-size-fits-all approach is not possible, prospective randomised controlled trials are lacking, and most surgical recommendations rest on level 3 or level 4 evidence. [2]

Sizing the airway and comparing the narrowing against expected anatomical norms is one of the most critical steps in evaluation, and it is done with endotracheal tubes during microlaryngobronchoscopy against a published sizing chart. The infant larynx differs from the adult: it sits higher in the neck, with the hyoid bone overlying its most superior aspect. [2]

The narrowest part of the infant airway is the complete cricoid cartilaginous ring — the only fixed ring within the airway, and therefore the site of most iatrogenic damage caused by intubation. That single anatomical fact explains why the subglottis is where intubation injury lands, and why expanding a complete ring is harder than expanding anything else. [2]

The supraglottis is bounded superiorly by the superior edge of the epiglottis, the aryepiglottic folds and the arytenoids, and inferiorly at the level of the true vocal folds. The glottis comprises the true vocal folds and the glottic chink posteriorly, and in the infant the posterior 50% of the true vocal fold is made up of the vocal process of the arytenoid. [2]

The subglottis sits between the undersurface of the vocal folds and the lower border of the cricoid cartilage. Its mucosa is lined with respiratory epithelium, with a transition to squamous epithelium at the free edge of the true vocal fold — a boundary worth knowing when interpreting where scar and where normal mucosa should lie. [2]

Laryngofissure with dermal grafting for adult scar tissue dates to 1938, when the leading cause of subglottic stenosis was high tracheostomy performed as an emergency for diphtheria. Splitting of the cricoid ring followed in Budapest in 1953, when Rethi described anterior and posterior splitting to treat cicatricial stenosis in adults — largely for war-related laryngeal injury — leaving a long-term rubber stent for months before wound closure. [2]

In 1966 a stent was introduced which could be wired into the tracheostomy tube and left for several months, and the first description of the Rethi procedure in children followed in 1971, with the incision closed entirely using an Aboulker stent wired to a metal tracheostomy tube. The modern era began in 1972, when Fearon and Cotton in Toronto reported a cartilage graft positioned in an anterior cricoid split to maintain expansion of the ring. [2]

That report included early reconstruction using auricular cartilage with skin attached and rib cartilage lined with buccal mucosa, and the African green monkey proved a helpful model using a trapdoor technique based on thyroid cartilage. Variations of those early techniques are still used, although graft lining is now better achieved with the inner perichondrium and the skin removed. [3]

Laryngotracheoplasty followed in 1974: trachea and cricoid split using a castellated incision and reapproximated around a rolled silastic sheet, which achieved successful cricoid expansion. Combined with reconstruction it was the primary procedure of the 1970s and 1980s, and even in those early days the decannulation rate for cartilage grafting was around 90%. [3]

The next innovation was cartilage grafting into the posterior cricoid lamina. Animal data showed posterior grafts achieved decannulation with significantly less absorption and necrosis than anteriorly placed grafts. Combining anterior and posterior grafts and adding lateral cuts to form the four-quadrant split meant almost any subglottic stenosis could be managed surgically — though severe disease still did less well. [3]

Where the principal management was tracheostomy, giving the child a safe airway while they grew out of the stenosis, the majority of children did not outgrow the problem. In 1977 Cotton performed the first anterior cricoid split to facilitate extubation and it succeeded; he published initial results in 1980, and a further series in 1988 of 67 patients reported a 70% successful decannulation rate. [3]

The split is no longer commonly used as an isolated procedure, but it retains a role in children with early stenosis who fail extubation and have no baseline oxygen requirement. Its wider importance was conceptual: it demonstrated that laryngotracheal surgery could be performed without a tracheostomy at all, which gave rise to single-stage reconstruction in the 1990s with the tracheostomy removed during surgery and the patient intubated for 2 weeks. [3]

Decannulation rates following laryngotracheal reconstruction are reported at over 90%, and even revision cases succeed at around 90% — a figure worth carrying into a conversation about a failed first operation. It is nonetheless essential to wait at least 6 months after the initial reconstruction before performing a revision. [3]

Resection took longer to reach children. Conley reported successful resection with primary end-to-end anastomosis in adults in 1953, but there was initial reluctance in children from concerns about growth and about the stenosis commonly merging into the glottis or trachea. A modification was performed on a 14-year-old boy in 1974, and Monnier published the first successful paediatric series in Switzerland in the 1990s, decannulating 36 of 38 children. [3]

Children present variably — acute airway compromise, stridor, shortness of breath or increased work of breathing — or are referred after failure to intubate or extubate, or arrive with a tracheostomy on which they are entirely dependent. Whatever the route, careful history from the child where age allows, the family, the referrer and collateral sources is vital, and the first task is to determine the acuity and plan for airway safety and stability. [3]

Where the patient is stable, take a perinatal history, a family history including associated syndromes, and the relevant medical history. Four specific things must be asked about because each changes the operative decision later: a past diagnosis of gastro-oesophageal reflux, cardiac disease, lung disease and aspiration. [3]

For stridor, establish duration. Acute stridor is a potential airway emergency; for chronic stridor determine the mode of onset and the nature, including whether it is stable or progressive. Assess impact through shortness of breath on exertion, increased work of breathing, exercise intolerance, failure to thrive and implications for sleep — noting that obstructive symptoms in sleep are more likely secondary to adenotonsillar hyperplasia than to supraglottic pathology. [3]

The intubation history is where the aetiology usually lies: duration, number of intubations, whether an age-appropriate tube was used, and the timing of onset of symptoms after intubation. In a child who is currently intubated, take the events requiring intubation and the ease or unease of it, along with tube size and type including whether cuffed, time intubated, number of attempts, ventilation settings, air leak, and any condition that may cause extubation to fail. [3]

In the tracheostomy-dependent child, establish the aetiology that required the initial placement, the duration of cannulation and the size of the tube. Any history of airway reconstructive surgery is essential, as is a thorough evaluation of voice quality and of the ability to tolerate a speaking valve — the latter says a great deal about the airway above the tube. [3]

Assess clinical signs of airway distress carefully: stridor, tracheal tug, and intercostal and sternal recessions. Characterise stridor by the phase of breathing, expiratory, inspiratory or biphasic. But retractions are the better indicator of the severity of obstruction — and the warning that follows is the one to remember, to beware the quiet child with significant retractions and work of breathing. [3] [4]

Voice quality localises the lesion. Children with subglottic and posterior glottic stenosis generally have a normal voice. Supraglottic stenosis often produces a muffled voice, and anterior glottic webbing tends to produce a hoarse voice. That triad lets a level be predicted before any instrument is passed. [4]

Flexible nasal endoscopy gives useful information in stable children but is in no way a replacement for microlaryngobronchoscopy in theatre, since little beyond the supraglottis and glottis can be assessed. Its value is in specific questions: evaluating other causes of obstruction such as nasal obstruction and glossoptosis in a child who has failed extubation, and assessing the position of a tracheostomy tube and whether granulation tissue or mucus plugging is present. [4]

The most critical single piece of information from flexible endoscopy is vocal cord movement, though choanal atresia, laryngomalacia and adenotonsillar hyperplasia can also be assessed. One hard rule governs it: the scope should not be passed below the level of the glottis, because this can induce laryngospasm. [4]

Imaging is not routinely performed on every child with airway symptoms, and many reach the operating room for microlaryngobronchoscopy without any. Where it is used, soft-tissue lateral and anterior-posterior neck and chest X-rays help evaluate laryngeal and tracheal stenosis, and can detect mediastinal pathology causing external compression or lung pathology causing respiratory distress. [4]

A barium swallow is an instrumental investigation here, giving information about aspiration risk, propensity for reflux, oesophageal foreign body and vascular compression, and it may indicate the need to evaluate the airway for tracheo-oesophageal fistula and laryngeal cleft. Endoscopic airway evaluation is commonly performed with a bronchogram where stenosis or tracheobronchomalacia is suspected. [4]

Spiral CT with contrast enhancement gives a helpful view of intrathoracic vasculature, and 3D reconstruction of the axial images may be valuable in evaluating the intrathoracic airway where relevant. But the governing statement is the last one: there is no radiological equivalent to endoscopic airway evaluation, so imaging supplements the scope and never replaces it. [4]

Flexible bronchoscopy under general anaesthesia may complement rigid assessment and in many instances the two are interchangeable. Rigid bronchoscopy provides superior laryngeal assessment; flexible sees around corners and reaches areas the rigid scope cannot, including beyond the eighth generation of bronchi in older children. Flexible is also what allows dynamic assessment such as drug-induced sleep endoscopy, and lavage detecting lipid-laden macrophages identifies silent aspiration. [4]

Rigid endoscopy under general anaesthesia is the gold standard for paediatric airway assessment, and the set-up matters as much as the technique. Assembling an expert surgical, anaesthetic and nursing team with appropriate equipment is critical, with a careful surgical brief and time-out, identified roles for every team member, and an explicit discussion of how the deteriorating patient will be managed. [4]

Preoperative dexamethasone at 0.5 mg/kg, up to a maximum of 20 mg, is necessary in all patients without a tracheostomy. Excellent communication with the anaesthetist is crucial because these are shared airway cases. The options are total intravenous anaesthesia, jet ventilation, assisted ventilation, apnoea with intermittent bag-mask ventilation, and spontaneous ventilation — the last preferred for its superior dynamic airway assessment. [4]

One workable technique is the child spontaneously ventilating on a mixture of sevoflurane and oxygen through an endotracheal tube placed in the oropharynx, with additional intravenous anaesthesia by propofol bolus. Endoscopy may use the Hopkins rod telescope with a straight or curved laryngoscope blade, a ventilating bronchoscope, or a suspended laryngoscope which frees both of the surgeon's hands. The technique matters less than the information gained. [4]

The assessment runs from outside in. Begin by looking for retrognathia, macroglossia, glossoptosis, tonsillar hyperplasia, pharyngeal scar and a difficult laryngeal view. Then the supraglottic larynx for laryngomalacia, scarring, omega-shaped epiglottis, short aryepiglottic folds and arytenoid prolapse. Then the glottis for webbing, posterior stenosis, scar, laryngeal cleft and vocal cord movement. [4]

Where subglottic stenosis is found, record position, length, grade and characteristics — whether soft or firm, concentric or with lateral shelving, and whether the mucosa is quiescent or actively inflamed. Those characteristics decide whether an endoscopic approach is even worth attempting. Hopkins rod endoscopes with an outer diameter as small as 1.9 mm can negotiate most grade III stenosis. [4] [5]

Assessment of the upper trachea covers the tracheostomy stoma site, looking for suprastomal collapse or suprastomal granuloma, and then possible tracheomalacia, vascular compression, the carina and the mainstem bronchi. Complete tracheal rings are rare but their possible presence must be evaluated cautiously, so as not to induce oedema within an already compromised airway segment. [5]

Airway reconstructive surgery is not always required and the decision should be made in a multidisciplinary setting. The governing principle is stated plainly: it is not usually the degree of stenosis but the child's overall health status that determines whether to proceed. Improving the airway can carry a significant trade-off against voice and swallowing, and there is a delicate balance to strike. [5]

Three situations argue against reconstruction. As children grow, mild stenosis becomes less of a problem with time. For children who chronically aspirate, laryngotracheal separation may be more appropriate than reconstruction. And in children requiring positive pressure ventilation, reconstruction is unwise — the operation cannot deliver what it exists to deliver. [5]

Three things should be optimised first. Gastro-oesophageal reflux disease is thought to contribute negatively to reconstructive outcomes and should be managed first, though this remains slightly controversial. Screening for oxacillin or methicillin-resistant Staphylococcus aureus before open airway surgery is recommended, as postoperative infection with these organisms may compromise the repair. Weight matters too. [5]

The weight and maturity thresholds are specific: children weighing more than 4 kg and those with a gestational age greater than 30 weeks appear to have a greater chance of successful extubation and an eventual patent airway. Because so many of these children now carry syndromes or extreme prematurity, most are discussed in a multidisciplinary setting and optimised preoperatively with expert advice from several paediatric subspecialists. [5]

In some children, observation before surgical intervention is the best course, particularly where the larynx is active or inflamed — usually secondary to reflux. One rule follows from this and should not be broken: reconstruction should not be performed if the aetiology of the laryngeal inflammation is unknown, because operating into inflammation is what produces recalcitrant restenosis. [5]

Azithromycin is given as an anti-inflammatory drug on a Monday, Wednesday and Friday dosing regimen, and may effectively reduce inflammation in up to 50% of children with an idiopathic active larynx. In children who do not respond to it, spontaneous resolution often occurs within a 1- to 2-year period — which makes waiting a real option rather than a failure to act. [5]

Before resection or reconstruction it is important to exclude concomitant oesophagitis, reflux and gastritis, to assess patency of the upper and lower oesophageal sphincters, and to determine the function of any previously performed fundoplication. This is achieved with oesophagogastroduodenoscopy, oesophageal biopsy and pH probes, which can be placed at the same time as endoscopy or even during airway evaluation. [5]

Two practical points govern the medication. Patients should be off antireflux medication for at least 1 week before pH probe study. Impedance testing also evaluates non-acid reflux, so patients can often stay on antireflux medicines during that study — a difference worth knowing before cancelling a child's treatment unnecessarily. [5]

Preoperative reflux evaluation is vital in children with recalcitrant stenosis after a previous reconstruction, and in those with an active larynx. That evaluation uses impedance monitoring, dual probe pH and oesophageal biopsy, and the biopsies are not optional: they are essential in diagnosing eosinophilic oesophagitis, which is a distinct and treatable cause of a hostile larynx. [5]

Aspiration is the enemy of the paediatric airway surgeon, because while improving the airway the risk of turning a child into a chronic aspirator is inherent in what the operation does. A preoperative evaluation should always be performed with a barium swallow or video swallow or both, with the local speech and language therapist reviewing the patient alongside the surgeon. [5]

Even where a child has a normal study from another centre, the images should be checked and the study repeated if they are not accessible — not all centres are expert in evaluating paediatric swallow studies, and the problem is compounded by there being no normative paediatric data at all. That absence is why the surgeon's own reading of the images matters. [5]

Clinically, ask about recurrent chest infections and drops in baseline oxygen saturations, which are highly suggestive of chronic aspiration. In children with tracheostomy tubes, secretions stained with food material are indicative. Two assumptions must be discarded: a cuffed tracheostomy tube does not prevent aspiration, and children with gastrostomy tubes can still reflux and can still be at risk. [5]

Several tests are available. A chest CT may show infection or parenchymal changes consistent with chronic aspiration. A food dye test places blue or green dye on the tongue and suctions tracheal secretions to see whether coloured secretions appear. A radionucleotide spit study is another option but is uncommonly performed, and bronchoalveolar lavage aspirate can be sent for lipid-laden macrophages. [5]

The grade IV child is a particular challenge, because with a completely obstructed airway no frank aspiration will be seen. Laryngeal penetration increases the risk and should be considered preoperatively in that setting. Functional endoscopic evaluation of swallowing, visualising the larynx with a nasopharyngoscope while the child swallows, gives the risk and mechanism of aspiration and shows whether laryngeal sensation is normal. [5] [6]

Where the risk of aspiration is high, children should undergo minimisation procedures before airway surgery is contemplated. These are conversion to gastrostomy tube feeding, reduction of salivary production, and performing a fundoplication — three interventions aimed at the swallow rather than the airway, done so the airway operation does not make a chronic aspirator. [6]

The primary purpose of reconstruction is to enable decannulation, and if that is not achievable there is little point in considering it. Children reliant on long-term ventilation or continuous positive airway pressure are unlikely to do well. Those with central hypoventilation syndrome, diaphragmatic weakness and progressive neuromuscular disorders are unlikely candidates for decannulation, so a formal sleep study benefits preoperative planning and results should be discussed with respiratory physicians. [6]

Reconstructive surgery may impair the voice and affect quality of life, with voice quality afterwards varying from normal to moderately impaired. One counter-intuitive finding is worth carrying into consent: the degree of dysphonia does not always correlate with voice-related quality of life scores, so a rough-sounding voice is not automatically a distressing one. [6]

Where a laryngofissure is required to reconstruct the airway, accurate height reapproximation must be achieved to avoid adverse voice outcomes. And for the most severe group the calculus inverts: children with grade IV stenosis are aphonic unless oesophageal speech has been employed, so reconstruction may give them their best chance of obtaining speech at all, even where decannulation is not achieved. [6]

Medical therapy is seldom used as primary treatment for airway stenosis, but it plays a critical role in stabilising the active or inflamed larynx and creating a suitable environment for surgery. In some circumstances it may prevent the need for reconstruction entirely, by giving sufficient improvement to permit decannulation. It comprises proton pump inhibitors or H2 antagonists for reflux, and azithromycin prophylaxis for the active larynx. [6]

In many countries the primary treatment is tracheostomy, staying in place until decannulation can be attempted when the child is older, or until reconstruction can be performed. A tracheostomy is a safe airway only where appropriate training and community supervision are available — a qualification that determines whether it is a solution or a hazard in a given setting. [6]

The level of the tracheostomy cuts both ways. A high tracheostomy increases the risk of worsening the stenosis, but it can make cricotracheal resection easier by leaving a smaller segment to resect. Set against any of this is the plainest statement in the section: a tracheostomy tube is a marked compromise on both the child's and the family's quality of life and emotional health. [6]

Endoscopic techniques can improve stenosis before reconstruction, serve as a primary method, or act as adjuvant treatment before or after open surgery. The general rule for choosing between them is grade-based: more severe grade III and IV stenosis requires open techniques, while grades I and II are amenable to endoscopic methods. [6]

Balloon dilation is the workhorse of endoscopic technique, widely adopted over the past 15 years. Early dilation, with the mucosal trauma caused by the shear forces inevitable in that technique, has fallen from favour and been replaced by high-pressure non-compliant balloons. Balloons have reportedly reduced the need for open airway surgery by up to 80%. [6]

The features that make a balloon work are worth knowing because they explain the shift. A balloon exerts a purely radial force over the circumference of the stenosis, which minimises the risk of mucosal trauma and airway rupture. Pressures of over 20 atmospheres can be achieved, guided by the surgeon with a handheld gauge, and the catheters are narrow and flexible enough to glide through even severe stenosis. [6]

Sizing follows the expected normal airway diameter rather than the stenosis, and for tracheal lesions the size chosen is typically 1 mm greater than that predicted for the larynx. Repeated dilations at 2- to 3-week intervals for around four treatments are recommended. Endoscopic adjuncts to dilation are scar division with cold steel or laser, steroid injection, and topical application of mitomycin C. [6]

The best candidates are children with soft, thin stenosis, though patients with firm mature scar tissue can still benefit from scar division combined with dilation. The patient must not breathe forcefully during dilation; since these cases are often part of a microlaryngobronchoscopy the child is not likely to be paralysed, so ask the anaesthetist for a propofol bolus to minimise spontaneous respiration. [6] [7]

Pre-oxygenate before inflation. Introduce the balloon through the stenotic segment and inflate to the required pressure, pulling it back towards yourself so it stays in place rather than migrating distally. Inflate ideally for 2 minutes, or until desaturation falls below 90%, then deflate completely and remove. Pre-oxygenate again before repeating up to three times, using a range of balloon sizes as the stenosis expands. [7]

Complications are uncommon but specific, and five are named. Airway rupture. Negative pressure pulmonary oedema. Distal displacement of the balloon, known as watermelon seeding. Failure of the balloon to deflate, which requires bursting it. And failure to improve the stenosis at all — in which case open surgery should be considered rather than repeated attempts. [7]

There is no single operation that addresses all airway stenosis: every child varies in the severity, location and length of the stenosis, and many carry medical comorbidities that add complexity. The mainstay of laryngotracheal reconstruction is expansion cartilage grafting. In the subglottis the alternative is resection and reanastomosis; in the supraglottis good results can be achieved with laryngoplasty without any cartilage graft; and tracheal stenosis is best managed with a slide tracheoplasty, usually transthoracic though certain cases can be transcervical. [7]

Costal cartilage is the most common graft — readily available, easily shaped and carved, harvested between the right fifth and sixth ribs, at the inframammary crease in girls, and more than one graft can be raised from it. One technical rule governs its use: the perichondrium should be left intact and must face the lumen of the airway when the graft is secured. [7]

Thyroid ala is useful for small grafts, taken from the upper aspect of the thyroid cartilage at least 1 mm above the vocal cord, and it has the advantage of being harvested from within the surgical field. Auricular cartilage is abundant and easy to harvest and useful in suprastomal collapse in a single-stage procedure, but it is weaker than the alternatives and unsuitable to support the cricoid split. [7]

Materials described but not routinely used are buccal mucosa, septal cartilage, hyoid bone and clavicle. In practice the choice comes down to costal cartilage where the graft must support the cricoid split, thyroid ala where a small graft will do and a second incision is not wanted, and auricular cartilage for suprastomal collapse. [7]

At the laryngeal level, stents maintain patency after two-staged reconstruction; in a single-stage procedure the stent can only be set while the endotracheal tube is in place. The Montgomery stent is made to fit into the larynx and is usually placed via a laryngofissure, made of silastic — which does encourage granulation, and that is the trade-off it carries. [7]

The Aboulker stent is a Teflon tube with a dome at the top to prevent aspiration; where it sits with the tracheostomy tube passing through it, a hole must be made in the stent and the tracheostomy tube wired onto it. Teflon tends to be associated with less granulation than silastic, which is the reason to reach for it. [7]

The LT-Mold is a newer stent developed from studies using cadaveric larynges, whose anatomical likeness is thought to produce a more acute angle at the anterior commissure after reconstruction. Where it is unavailable, a similar technique has been described using hearing aid silicone to create a customised laryngeal stent — a genuinely useful improvisation. [7]

Rutter stents are soft and deformable, do not interfere with tracheostomy tube placement, incite less granulation at the distal end, and leave less dead space between stent and tracheotomy tube in which scarring can occur. The proximal end is trimmed to the level of the false vocal cord and plugged with a rounded cap less likely to induce epiglottic granulation. [7]

Unlike the rigid Aboulker Teflon stents, these soft silastic suprastomal stents may remain in place for over 6 weeks — with a corresponding speech restriction, since they permit only minimal air passage. That is the counselling point when a child is going to carry one for a prolonged period. [7]

Stents also treat stenosis in the subglottis and trachea, deployed under direct vision. The significant problem has been granulation tissue and in-stent stenosis, sometimes making removal nearly impossible, mainly with nickel-titanium stents. Those results led the US Food and Drug Administration to recommend against using stents in benign laryngotracheal stenosis in 2005. [7]

Drug-eluting stents may reduce the granulation problem: coated with medications such as rapamycin, they carry a second advantage in being made of resorbable materials that do not need removal. Animal trials are still underway and no human data are published. The Dumon stent is silicone with external studs preventing migration, though granulation at both ends remains a concern and mucus plugging can occur. Expandable stents made of dissolvable Vicryl are also available. [7] [8]

T tubes were introduced in 1965 by William Montgomery and are silicone tubes in the shape of a T. The vertical component has proximal and distal flanges with an external horizontal arm traversing the tracheostomy stoma, a design that facilitates suctioning of secretions and maintenance of a patent airway. Because of the dual flange system, air can still pass through one flange if the other becomes blocked. [8]

They are used most commonly for stenosis in the subglottis and cervical trachea, and can be placed directly into the airway during open reconstruction, or endoscopically though that is more difficult. Their distinctive advantages are that they allow phonation without a speaking valve by occluding the outer flange, and that they uniquely allow ventilation through the tracheostomy site while stenting the proximal airway. [8]

The proximal end can be varied and may extend through the glottis to stent a large segment of the upper airway. The biggest problems are mucus plugging and, in children, decannulation caused by pulling on the outer flange. Bacterial biofilms have been associated with all forms of stent, with organisms such as Staphylococcus and Pseudomonas species. [8]

To overcome blunting of the anterior commissure, a spring to treat posterior glottic stenosis can be created using a T tube. Cadaveric studies showed the pressure from the spring does not disrupt the cricoarytenoid joint capsule, and a subsequent study of five patients with posterior glottic stenosis treated with the posterior glottic spring saw all of them successfully decannulated. [8]

An inter-arytenoid spring using a modified safety pin exists only in a porcine model, and it was developed to treat bilateral vocal fold paralysis rather than posterior glottic stenosis — a distinction worth preserving when reading about it. For glottic stenosis from webs or bilateral vocal cord palsy, the Lichtenberger needle holder allows endolaryngeal suturing, or an 18 gauge needle can be passed externally through the cricothyroid and thyrohyoid membranes. [8]

Keels prevent formation of an anterior glottic web where both sides of the anterior commissure are involved during surgery. Other adjuncts for posterior glottic stenosis include a temporary silastic sheet sewn around the scar band between the vocal cords, allowing mucosalisation before the final scar band is divided. Extralaryngeal bioabsorbable mini plates are useful in suprastomal collapse but are not yet commonly used. [8]

Single-stage surgery can be considered where less than 2 weeks of stenting is anticipated, the endotracheal tube acting as the stent. Beyond 2 weeks there is a risk of redeveloping subglottic or posterior glottic stenosis. Optimal timing of intubation afterwards is unknown, but the literature favours 2–7 days for resection and anterior grafts, and 7–14 days for anterior and posterior grafts, varying with age and complexity. [8]

Success depends on the unit as much as the operation: the institution must have an excellent paediatric intensive care unit familiar with the procedure. Security of the nasotracheal tube is paramount and all staff must understand this. Sedation and arm restraints are usually required in children under 3 years of age, and single-stage procedures should not be done in children who are difficult to intubate, in case of accidental extubation. [8]

If not sedated, children do not require ventilatory support and can often ambulate. On the day before extubation it is standard practice to take the child to theatre to inspect the airway and downsize the endotracheal tube; if an adequate airway is seen, a dose of dexamethasone is given and the child extubated the next day during daytime hours with appropriate staff available. [8]

In airway distress after extubation, nebulised adrenaline, further dexamethasone, heliox or positive pressure ventilation can all be tried before reintubating. If re-intubation occurs, allow a few more days before another trial of extubation; if that fails and a tracheostomy is required, it should be placed below the graft site so the reconstruction is not compromised. [8]

Patients best suited to single-stage reconstruction require a simple procedure and have a stable larynx. Those who should not undergo it are children who are difficult to intubate, those with poor pulmonary function, those who have failed a previous reconstruction, and those with complex disease involving multiple levels of obstruction. [8]

Supraglottic stenosis is rare and associated with supraglottic collapse, and its primary complaint is obstructed breathing during sleep rather than stridor. A sleep study is beneficial in patients who are not tracheostomy tube dependent. The underlying aetiology is often traumatic or iatrogenic, from several previous laryngeal reconstructions. [8]

Flexible endoscopic evaluation with the patient spontaneously ventilating allows the dynamic assessment this diagnosis needs. Nocturnal continuous positive airway pressure may be very beneficial in mild to moderate supraglottic stenosis and collapse. Arytenoid prolapse is usually managed endoscopically with a partial arytenoidectomy — raising a mucosal flap and debulking the prolapsing cartilage without damaging the mucosal diamond of the laryngeal inlet. [8]

The petiole is the base of the laryngeal surface of the epiglottis; when it prolapses it obscures the anterior true vocal cords and shortens the anterior-posterior diameter of the laryngeal inlet. It is rare, most commonly seen in children who have had repeated laryngofissures, and is a consequence of damage to the thyroepiglottic ligament. [8] [9]

Petiole prolapse is challenging to treat. Suspension of the epiglottic base to the hyoid bone provides some benefit, but optimal management is a complete laryngofissure repositioning the petiole to the inner surface of the thyroid ala. The laryngofissure is then closed over a T tube or suprastomal stent, left in position as a translaryngeal stent for at least 2 months. [9]

Glottic webs are most commonly congenital and usually associated with subglottic stenosis; acquired webs are less common and post-traumatic, from anterior neck trauma or iatrogenic injury. The two are managed differently for a reason that is purely histological: mucosa is normal in congenital webs, so there is often enough mobility to reconstruct the anterior commissure without a keel. [9]

Acquired webs are associated with fibrosis and scarring, so reconstruction with placement of a laryngeal keel is mandatory while the raw surfaces on either side of the laryngofissure heal. Repair uses an open approach with complete laryngofissure, and a pexing suture may be placed from the cut mucosal edges on either side towards the thyroid ala — though it is often not possible to pex those edges up to the anterior commissure, which is precisely why the keel is needed. [9]

The keel is selected and trimmed so its vertical height separates the raw surfaces of the laryngofissure, with the upper limit not so high as to disrupt the insertion of the petiole. It is sewn into place and the laryngofissure closed, often as a two-stage procedure, though it can be done single-stage with the endotracheal tube positioned on one side of the vertical limb of the keel. [9]

Where subglottic stenosis is associated, decide whether the cricoid can be adequately closed over an age-appropriate endotracheal tube; if not, an anterior cartilage graft is positioned in the anterior cricoid, distal to the keel. The keel is removed via an open approach between 10 days and 4 weeks postoperatively, the midline deficit in the thyroid ala closed with laterally placed mattress sutures, with antibiotic coverage and antireflux measures while it is in place and for a few days after removal. [9]

Endoscopic keel placement is gaining popularity. The larynx is suspended and the web divided endoscopically with a sickle knife, and the keel made from a thin piece of silastic sheeting with a central suture orientated in the anterior midline, the sheet placed like the leaves of a book over the divided mucosa. The suture is placed either inside-out with a Lichtenberger needle driver or outside-in using a Keith needle and hollow angiocath. [9]

Posterior glottic stenosis is frequently misdiagnosed and confused with bilateral vocal cord paralysis, which is the single most important thing to know about it. The most frequent aetiology is prolonged intubation. Patients present with stridor and exertional dyspnoea, and normal vocal function is usually preserved — so a normal voice does not exclude a significant airway problem here. [9]

Diagnosis requires assessment with rigid bronchoscopy, and evaluation should include a review of the subglottis, which is frequently involved with scarring. Arytenoid mobility must be assessed because cricoarytenoid joint fixation is an occasional co-pathology. The distinguishing finding is that the vocal folds are mobile but tethered and unable to abduct; true bilateral paralysis and posterior glottic stenosis rarely coexist. [9]

A posterior costal chondral graft is the mainstay of management and a highly effective way of achieving an adequate glottic airway, best performed through an anterior approach. The posterior glottis is infiltrated with 1% lignocaine with adrenaline, then the posterior cricoid split vertically for its entire length, keeping the incision entirely in the midline, and the interarytenoid scar band completely divided. [9]

Following the split the cricoid should be easily distracted laterally. A costal graft is carved so that its height approximates the height of the cricoid split and its depth allows the graft perichondrium to lie reasonably flush with the cut mucosa of the posterior cricoid. It may be sewn in place with a 4.0 Monocryl suture, or formed as a flanged graft snapped into place and stabilised with a small amount of fibrin tissue glue. [9]

With the posterior graft in place the laryngofissure is closed over an age-appropriate endotracheal tube, and if the cricoid does not close anteriorly a further segment of costal cartilage is used as an anterior graft. A posterior graft rarely needs to be more than 6 mm wide, because the primary aim is to incise the scar tissue and hold the raw edges apart while healing takes place — not to create a large lumen mechanically. [9]

Posterior cricoid grafting may be single- or two-stage, and appropriately placed grafts correct the subglottic stenosis and can slightly separate the anterior commissure to prevent webbing. The endoscopic route has become increasingly popular over the last 10 years: it requires no anterior component of subglottic stenosis and a larynx that can be adequately exposed. [9]

Endoscopically, the posterior cricoid may be split with a laser or a sickle knife and balloon dilation used to distract the split. Pockets are created behind the posterior plate of the cricoid to accommodate the graft's posterior flanges; significant force is required to place the graft, and balloon dilation may assist. Other endoscopic options described are microtrap-door flaps, mucosal advancement flaps, vocal cord lateralisation and botulinum toxin injection. [9]

The criteria for an anterior cricoid split are strict. Failed extubation on at least two occasions, weight over 1500 g, and extubation failure secondary to laryngeal pathology. No assisted ventilation for 10 days before the evaluation, supplemental oxygen requirement under 35%, and no antihypertensive medication for 10 days before evaluation. No congestive heart failure for 1 month before evaluation, and no acute upper or lower respiratory tract infection at the time. [10]

The procedure incises the trachea anteriorly from the second tracheal ring, up through the cricoid and into the lower third of the thyroid cartilage, below the insertion of the anterior commissure. The child is left intubated for 7–10 days and the neck wound left partly open to minimise the risk of subcutaneous emphysema. Earlier extubation can be achieved with a thyroid alar interposition graft. [9] [10]

Mild to moderate subglottic stenosis is well managed with costal cartilage grafting to the anterior cricoid. The anterior airway is split from the tracheotomy site to the lower aspect of the thyroid cartilage, an age-appropriate endotracheal tube or suprastomal stent inserted, and the graft size measured to close the deficit comfortably. [10]

The graft is carved as a boat-shaped, perichondrium-lined insert that distracts the anterior cricoid, with an outer flange preventing the graft prolapsing into the airway. The same technique helps in managing suprastomal collapse or narrowing of the upper trachea. Posterior cricoid grafting for subglottic stenosis is performed identically to that for posterior glottic stenosis. [10]

Additional anterior grafting is not required where the anterior cricoid can close comfortably over an appropriately sized endotracheal tube or stent. Where it cannot, an additional anterior graft is required — and that combination is necessary for most grade III and all grade IV stenoses, which is the practical meaning of the grade in the operating theatre. [10]

Immediately before the procedure a rigid endoscopy confirms the exact nature and position of the stenosis, and an oesophageal bougie is inserted to help define the position of the oesophagus during the later tracheal mobilisation. The neck is extended initially, but returned to the anatomical position when the anastomosis is performed, so that tension is reduced. [11]

Where a tracheostomy is present, an oral RAE tube is cut short and placed in the stoma to allow ventilation, a large size chosen to reduce the intraoperative air leak once the airway is opened. If a single-stage procedure is planned an oral endotracheal tube is placed instead. Local anaesthetic with adrenaline is infiltrated around the tracheostomy and an elliptical incision made around the stoma. [11]

Superior and inferior subplatysmal flaps are elevated to the hyoid bone and the sternal notch, and the strap muscles divided in the midline to expose the airway. Care is taken to preserve the cricothyroid muscle, which serves as a valuable landmark identifying the cricoid and therefore the level of the stenosis — losing it loses the orientation the rest of the operation depends on. [11]

A vertical midline incision through the cricoid cartilage is extended superiorly and inferiorly to identify the lumen and the stenotic area. In grade IV stenosis a repeat endoscopy is performed and a small pair of artery forceps used to break through the stenosis into the laryngeal lumen. It is not usually necessary to divide the anterior commissure. [11]

The upper resection margin is defined and extended laterally along the lower border of the thyroid cartilage towards the cricothyroid joint; anterior to the joint the incision turns inferiorly and the cricoid is divided vertically on both sides. Posteriorly the incision continues along the inferior border of the cricoid lamina, so the resulting superior end comprises thyroid cartilage anteriorly and cricoid lamina posteriorly, joined by the cricothyroid joints. [11]

The subglottic scar covering the inner surface of the cricoid lamina is excised, leaving an edge of healthy mucosa superiorly below the arytenoids. To reduce the prominence of cartilage posteriorly a drill is used — a 4 mm cutting burr then a diamond burr — to thin the cricoid lamina. The thyroid cartilage is then split anteriorly in the midline, which will accept the V-shaped notch of the lower segment when the anastomosis is formed. [11]

Mobilisation proceeds laterally and anteriorly, taking care not to injure the recurrent laryngeal nerves — which are reliably preserved by keeping the dissection close to the tracheal wall in the subperichondrial plane. Anteriorly the dissection continues deep to the innominate artery and down towards the carina, extended laterally with sharp dissection. [11]

Two 2/0 Prolene stay sutures are inserted into the tracheal wall laterally and used to elevate the trachea. Posterior dissection starts at the lower border of the cricoid, with the oesophageal bougie palpated to identify the oesophagus, and the trachea lifted on the retraction sutures and dissected free of it. The lower resection margin must contain a full-sized healthy tracheal ring to be confident of success. [11]

Where the tracheal tissue at the stoma site is necrotic or there is surrounding perichondritis, the stoma should be included in the resection. Before doing so a new tracheostomy is fashioned in healthy lower trachea and the cut oral RAE tube transferred to it to maintain ventilation. The first ring above the stoma is fashioned into a V shape to insert into the thyroid cartilage anteriorly. [11]

The posterior membranous trachea is fashioned into a tongue-shaped flap that will lie over the cricoid lamina. Laryngeal release is essential in reducing tension in the anastomosis: the hyoid is identified, an incision made along its superior surface, and the suprahyoid muscles dissected free of the hyoid bone in the subperiosteal plane using a Freers elevator. [11]

With the neck returned to the anatomical position, 2/0 Prolene sutures are placed posterolaterally in the trachea on either side and inserted into the thyroid cartilage anterior to the cricothyroid joint, approximating the ends while the posterior mucosal anastomosis is completed. The tongue-shaped flap of membranous trachea is sutured to healthy mucosa below the arytenoids and vocal cords using 4/0 Vicryl. [11]

A T tube or stent may be inserted at this point, with endoscopy confirming the upper end lies just above the level of the arytenoids and the lower end is well clear of the carina. The anastomosis is completed anteriorly with 3/0 Prolene and the V-shaped notch inserted into the midline of the thyroid cartilage — opening the superior end and reducing the size mismatch between the two ends. [11]

The strap muscles are closed over a Penrose drain, the subplatysmal layer with interrupted Vicryl, and the skin with a continuous subcuticular running Vicryl suture. In active children or after extensive resections the chin should be sutured to the chest for 1 week using 0 Prolene, to prevent neck extension disrupting the anastomosis. [11]

Slide tracheoplasty was created to expand a congenitally stenotic trachea due to complete tracheal rings, and may also be employed for acquired laryngotracheal stenosis. Although initially described as an intrathoracic approach, a cervical approach allows access to the upper two-thirds of the trachea, which brings it within reach without thoracotomy in many cases. [11]

It requires adequate exposure of the larynx and trachea, with the stenotic segment delineated by an assistant performing bronchoscopy while the surgeon places a 30G needle into the airway. Proximal and distal aspects of the stenosis are marked on the anterior airway and the length of the stenosis measured before any cut is made. [11]

The trachea then undergoes bevelled transection, commencing on the anterior trachea proximal to the midpoint of the stenosis and extending over two rings distally, with the posterior transection point at or just distal to the midpoint. The distal trachea is split posteriorly in the midline to just beyond the stenosis, and anteriorly the proximal trachea is divided beyond the stenosis, usually to the thyroid cartilage. The most stenotic segment may be resected. [11]

The trachea is anastomosed with a double-armed PDS suture using a running technique, with no attempt to keep the suture extraluminal, and care taken to tighten the running suture as the anastomosis is completed. A single proximal anterior knot completes it. The airway is leak-tested and then sealed with fibrin glue, and the patient is typically extubated at the end of the procedure. [11] [12]

Complications subdivide into intraoperative, early postoperative and late postoperative. Intraoperative complications are airway compromise with hypoxia, bleeding, pneumothorax and pulmonary hypertension — the first of these being the one the shared-airway set-up and the team brief exist to anticipate. [12]

Early postoperative complications are loss of the graft, infection, dehiscence of the anastomosis and air leakage from the operative site. In single-stage procedures there is the additional risk of accidental extubation, which is why security of the nasotracheal tube is stressed as paramount and why the difficult-to-intubate child is excluded from a single stage. [12]

The most significant long-term complication is failure of the reconstruction with restenosis of the subglottis, and the incidence of it is between 10% and 20% in most series. That figure belongs in every preoperative conversation, alongside the rule that revision should not be attempted within 6 months of the original operation. [12]

A surgeon facing a child with airway stenosis now has a wide range of management options, and the procedure must be tailored to the child as well as to the anatomy of the stenosis. Both cricotracheal resection and laryngotracheal reconstruction achieve high rates of decannulation, and there is a lack of prospective randomised controlled trials comparing the methods. [12]

What retrospective data support is a clean selection rule. Cricotracheal resection is the preferred option for grade IV and severe grade III stenosis that are clear of the vocal cords. As the less extensive procedure, laryngotracheal reconstruction is selected for grade II and less severe grade III stenosis. Paediatric airway reconstructive surgery remains a challenge whichever is chosen. [12]

Prevention of stenosis is better than cure, and non-invasive ventilation of premature infants can avoid intubation trauma altogether — which is continuous with what has already driven the tracheostomy rate from 20% down towards 1%. Separately, alternatives to autogenous cartilage for laryngotracheal reconstruction and airway repair may be derived from tissue engineering in future, removing the donor site from the operation. [12]

• Scott-Brown's Otorhinolaryngology and Head and Neck Surgery, 9th edn, Ch 144 Airway stenosis

  • [1] p.1541

  • [2] p.1542

  • [3] p.1544

  • [4] p.1545

  • [5] p.1546

  • [6] p.1547

  • [7] p.1548

  • [8] p.1549

  • [9] p.1550

  • [10] p.1551

  • [11] p.1552

  • [12] p.1553