Safety and efficacy of lingual tonsillectomy in multilevel airway surgery for pediatric obstructive sleep apnea

Abstract

Study Objectives:

Prior studies have demonstrated the efficacy of lingual tonsillectomy in treating pediatric obstructive sleep apnea. The goal of this study is to describe the postoperative outcomes following lingual tonsillectomy as a part of drug-induced sleep endoscopy-directed multilevel sleep surgery.

Methods:

A retrospective review was performed for pediatric patients with obstructive sleep apnea who underwent lingual tonsillectomy as a part of drug-induced sleep endoscopy-directed sleep surgery. Data collected included age, sex assigned at birth, body mass index z-score, polysomnography results, past medical and surgical history, and postoperative outcomes.

Results:

A total of 174 patients were included in the study with a mean age of 8.29 ± 3.49 years (range 1.89–15.62) and mean preoperative apnea-hypopnea index of 7.88 ± 13.42 (range 1.10–123.40). Complications occurred in 26 patients (14.9%) including 14 patients (8.0%) requiring emergency department visit or readmission and 12 patients (6.9%) experiencing postoperative bleeding. Asthma (P = .033) and developmental delay (P = .016) correlated with postoperative complications. For patients with preoperative and postoperative polysomnography data (n = 145; 83.3%), there was significant improvement (P < .001) in apnea-hypopnea index with a mean postoperative apnea-hypopnea index of 4.02 ± 7.81 (range 0.00–54.46). Surgical failure, defined as postoperative apnea-hypopnea index ≥ 5, was identified in 25 patients (17.2%). Surgical failure was associated with body mass index z-score > 2 (P = .025) and Trisomy 21 (P = .005).

Conclusions:

This study highlights the promising surgical success rate of drug-induced sleep endoscopy-directed lingual tonsillectomy in multilevel sleep surgery (82.8%) and infrequent complications including postoperative bleeding (6.9%) and readmission (2.3%).

Citation:

Williamson A, Morrow VR, Carr MM, Coutras SW. Safety and efficacy of lingual tonsillectomy in multilevel airway surgery for pediatric obstructive sleep apnea. J Clin Sleep Med. 2024;20(2):189–199.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Lingual tonsillectomy has been shown to be effective in the treatment of pediatric obstructive sleep apnea. However, there are few studies focused on postoperative complications and outcomes after lingual tonsillectomy in the context of drug-induced sleep endoscopy-directed multilevel airway surgery.

Study Impact: This study supports the implementation of drug-induced sleep endoscopy-directed lingual tonsillectomy in multilevel airway surgery for pediatric obstructive sleep apnea. These findings also identify comorbid conditions and concurrent procedures that may be associated with surgical failure and postoperative complications.

INTRODUCTION

Obstructive sleep apnea (OSA) is a condition with repetitive upper airway obstruction during sleep, characterized by hypopneas and apneas. OSA reportedly affects 1.2%–5% of children, although it is likely underdiagnosed and undertreated.^1–5 OSA in children is diagnosed by polysomnography (PSG) when the apnea-hypopnea index (AHI) is greater than one event per hour of sleep.^6,7 Adenotonsillectomy (AT) is first-line treatment recommended by the American Academy of Pediatrics and the American Academy of Otolaryngology Head and Neck Surgery for pediatric OSA.^7,8 Meta-analyses have demonstrated resolution of OSA following AT occurs in 59.8%–82.9% of patients.^9,10 Select populations are at an increased risk of persistent OSA following AT such as children with Trisomy 21, obesity, craniofacial syndromes, or severe baseline OSA.^11–13 Continuous positive airway pressure is one effective option for managing persistent OSA symptoms when patients have moderate (AHI ≥ 5) or severe OSA (AHI ≥ 10), including after failed AT.^14,15 Yet, guidelines regarding the need for repeat continuous positive airway pressure therapy titration testing and indications and timing for eventual cessation are not well defined.¹⁶ Continuous positive airway pressure is also not without risk of adverse events, as it has been associated with midface hypoplasia and retrusion as well as skin injuries.^17,18

Alternatively, further surgical intervention may be guided by drug-induced sleep endoscopy (DISE). DISE is an important tool to assist in surgical planning for children with persistent OSA after AT or in patients at risk for treatment failure following AT.^19–21 Several DISE scoring systems exist for classifying pediatric airway obstruction in order to guide therapeutic decision-making.^22,23 Tongue base and hypertrophied lingual tonsils are common sites of residual obstruction identified by DISE following AT.^22–25

Pediatric OSA has been successfully managed with DISE-directed lingual tonsillectomy (LT) in several studies using a range of surgical approaches including electrocautery, CO₂ laser, coblation, microdebrider, and transoral robotic surgery.^26–30 Although these studies have provided evidence to support the efficacy of LT in the treatment of OSA, the rate of and risk factors for postoperative complications following LT in DISE-directed multilevel sleep surgery has not been thoroughly studied.^31–33 Recent investigations have reported bleeding, decreased oral intake, and emergency department (ED) visit during the postoperative period of LT.^33,34 Others have highlighted rare respiratory complications postoperatively requiring interventions ranging from prolonged stay to advanced airway management.^35,36 In this study, we evaluated postoperative outcomes and complications of pediatric patients undergoing DISE-directed LT as a part of multilevel or single-level surgery for OSA. Specifically, length of stay, change in AHI and oxygen saturation nadir, and postoperative complications were studied.

METHODS

Study design and participants

A retrospective case series study was conducted at an academic tertiary care facility, following West Virginia University Institutional Review Board approval. A computational search for the term “lingual tonsillectomy” using the electronic medical record between January 1, 2013, and January 1, 2021, was performed. Patients under the age of 18 who had undergone LT for OSA were included. The database included sex assigned at birth, date of birth, past surgical history, past medical history, body mass index (BMI) z-scores, sleep study data, the indication for surgery, and the American Society of Anesthesiologists physical status classification. Comorbid medical conditions were collected from the medical record by using the International Classification of Diseases 10th Revision codification. Postoperative outcomes including length of stay, complications, postoperative bleeding, postoperative bleeding requiring operative intervention, poor oral intake, and prolonged dysphagia were stratified for analysis. Patients with BMI z-scores ≥ 1.645 (corresponding to the 95th percentile) of the same age and sex assigned at birth were considered obese.³⁷ The severity of sleep apnea was classified as mild (AHI ≥ 1 and < 5), moderate (AHI ≥ 5 and < 10), or severe (AHI ≥ 10) based on preoperative sleep study.

Surgical procedure and patient care

DISE was performed in the operating room using propofol sedation titrated and monitored by the anesthesia team following a standard protocol described in a previously published technical report.²³ DISE was interpreted using a previously published scoring system with good inter- and intrarater reliability.^23,38 Sleep endoscopy, DISE interpretation, and subsequent DISE-directed surgical interventions were performed by the senior author of the study. The predominant indication for DISE-directed LT was persistent OSA diagnosed by PSG following AT. Additional indications for DISE included patients considered to be at risk for persistent OSA following AT or those with small 1+ tonsils.^39,40 LT was indicated when DISE demonstrated hypertrophied lingual tonsil causing retrolingual airway obstruction.⁴⁰ Additional surgical interventions were performed concurrently with LT as indicated based on initial DISE findings including inferior turbinate reduction, adenoidectomy, midline posterior glossectomy (MPG), and supraglottoplasty.

For LT, general endotracheal anesthesia was employed by a pediatric anesthesiologist. With a dental guard in place, the pediatric Lindholm laryngoscope (Karl Stortz, Tuttlingen, Germany) was inserted and advanced to the vallecula to expose the lingual tonsil. Suspension was placed and the operating microscope was positioned for visualization. The EVAC 70 (Smith & Nephew, Watford, United Kingdom) coblation wand was then introduced alongside the laryngoscope and was used to ablate the lingual tonsils until the entire tongue base was cleared of all lymphoid tissue. Marcaine 0.25% with epinephrine 1:200,000 was injected into the right and left tongue base for postoperative local anesthesia to reduce the need for intravenous narcotic in the recovery room. The patient was then extubated and brought to the recovery room before transfer to a monitored bed for planned overnight observation.

Postoperative visits were scheduled approximately 1 month after surgery. Patients with moderate to severe OSA or persistent symptoms were advised to have postoperative PSG completed no sooner than 90 days after surgical intervention.⁴¹ For this study, surgical failure was defined as postoperative AHI ≥ 5. Patients were considered to have postoperative bleeding when documented on physical exam or, in cases of no exam findings, if the patient or parent reported oral bleeding. Patients with active bleeding or large volume bleed were taken to the operating room to isolate and control the source.

Statistical analysis

Statistical evaluation was performed using RStudio (Version 2022.12.0). Statistical significance was determined as P < .05. Patient subsets were compared using unpaired Wilcox rank sum analysis for continuous variables and chi-squared analysis for categorical variables. Postoperative PSG data was collected when available and used to calculate AHI and oxygen saturation improvement. Wilcoxon rank sum analysis was used to compare preoperative and postoperative AHI and oxygen saturation nadir. Multiple univariate linear and logistic regression models were compiled for each of the independent outcome variables including postoperative bleeding, postoperative readmission, or ED visit for poor oral intake or poor pain control. Additional univariate logistic regression models were compiled for the 145 patients with preoperative and postoperative PSG data available using surgical failure as the independent outcome variable. The estimated odds ratios (OR) and 95% confidence interval (CI) were calculated from the logistic regression models. Multivariate regression modeling was used for additional analysis of concurrent DISE-directed surgical procedures as independent variables. Standard deviation was used to demonstrate variation in descriptive statistics, and 95% CI was used to demonstrate variation in estimated OR from logistics regression modeling.

RESULTS

Participants

A total of 174 patients were included in this study. Of these patients, 98 (56.3%) were male, and 76 (43.6%) were female. The mean age of patients was 8.29 ± 3.49 years (range 1.89–15.62), and the mean BMI z-score was 1.01 ± 1.42 (range –3.69 to 4.22). The mean interval between preoperative PSG and procedure date was 5.5 ± 3.5 months. Additional clinical characteristics of the patients are summarized in Table 1. The most common comorbid conditions identified included gastroesophageal reflux (35.1%), asthma (35.1%), allergic rhinitis (34.4%), attention deficit hyperactivity disorder or another neuropsychiatric diagnosis (31.0%), and developmental delay (32.8%). The majority of patients underwent concurrent surgeries at the time of LT (161 patients, 92.5%).

Table 1.

Clinical characteristics of the study group.

Patient Characteristics	All Patients	Reported Complication	Postoperative Bleeding	ED Visit for Poor Oral Intake or Pain Control	Required Readmission for Poor Oral Intake or Pain Control
	(n = 174)	(n = 26)	(n = 12)	(n = 10)	(n = 4)
Age (years)
Mean ± SD	8.29 ± 3.49	8.87 ± 3.81	8.92 ± 3.81	8.63 ± 4.23	7.18 ± 3.33
Range	1.89–15.62	3.09–15.12	3.20–15.12	3.09–14.28	3.09–10.56
Age > 8 years, n (%)	84 (48.3)	13 (7.5)	6 (3.4)	5 (2.9)	2 (1.1)
BMI z-score
Mean ± SD	1.01 ± 1.42	1.31 ± 1.52	1.44 ± 1.85	1.22 ± 1.25	1.28 ± 1.72
Range	−3.69 to 4.22	−3.69 to 3.30	−3.69 to 3.30	−0.22 to 2.95	−0.22 to 2.95
BMI z-score > 1, n (%)	93 (53.4)	17 (9.8)	9 (5.2)	5 (2.9)	2 (1.1)
BMI z-score > 2, n (%)	55 (31.6)	10 (5.8)	6 (3.4)	3 (1.7)	2 (1.1)
Sex assigned at birth, n (%)
Male	98 (56.3)	18 (10.3)	9 (5.2)	6 (3.4)	3 (1.7)
Female	76 (43.6)	8 (4.6)	3 (1.7)	4 (2.3)	1 (0.6)
Surgical history, n (%)
Adenoidectomy	116 (66.7)	17 (9.7)	7 (4.0)	7 (4.0)	3 (1.7)
Tonsillectomy	99 (56.9)	15 (8.6)	7 (4.0)	6 (3.4)	2 (1.1)
ASA class, n (%)
ASA class I	2 (1.1)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
ASA class II	110 (63.2)	14 (8.0)	7 (4.0)	5 (2.9)	0 (0.0)
ASA class III	62 (35.6)	12 (6.9)	5 (2.9)	5 (2.9)	4 (2.3)
ASA class IV	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Comorbid conditions
Obesitya	72 (41.4)	12 (6.9)	7 (4.0)	4 (2.3)	2 (1.1)
Reflux	61 (35.1)	13 (7.5)	4 (2.3)	5 (2.9)	3 (1.7)
Asthma	61 (35.1)	14 (8.0)	5 (2.9)	7 (4.0)	4 (2.3)
Allergic rhinitis	60 (34.4)	9 (5.2)	2 (1.1)	5 (2.9)	2 (1.1)
ADHD / behavior	54 (31.0)	10 (5.8)	5 (2.9)	4 (2.3)	3 (1.7)
Developmental delay	57 (32.8)	14 (8.0)	7 (4.0)	5 (2.9)	3 (1.7)
Trisomy 21	10 (5.7)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Nonfebrile seizure	31 (17.8)	5 (2.9)	2 (1.1)	3 (1.7)	2 (1.1)

^a Obesity was defined as patients with BMI z-scores ≥ 1.645 (corresponding to 95th percentile) in accordance with the American Academy of Pediatrics. ADHD = attention deficit hyperactivity disorder, ASA = American Society of Anesthesiologists, behavior = other behavioral diagnosis, BMI = body mass index.

Postoperative complications

Complications occurred in 26 patients (14.9%) including postoperative bleeding (12 patients, 6.9%) and poor oral intake or uncontrolled pain prompting return to the ED (10 patients, 5.7%) (Table 2). One patient (0.6%) who underwent LT and adenoidectomy required reintubation immediately following surgery and was extubated following 3 days of steroid treatment for airway edema. There were 2 patients (1.1%) with prolonged dysphagia reported at the initial 1-month postoperative visit. This complaint had resolved by the second follow-up appointment, about 6 weeks postoperatively. One patient (0.6%) presented to the ED for globus sensation but was discharged with reassurance. Finally, 1 patient (0.6%) was also seen by their pediatrician postoperatively with fever and reported nasal drainage and was treated with an oral antibiotic.

Table 2.

Preoperative PSG and concurrent surgeries.

Patient Characteristics	All Patients	Reported Complication	Postoperative Bleeding	ED Visit for Poor Oral Intake or Pain Control	Required Readmission for Poor Oral Intake or Pain Control
	(n = 174)	(n = 26)	(n = 12)	(n = 10)	(n = 4)
Preoperative PSG study
AHI
Mean ± SD	7.88 ± 13.42	11.64 ± 24.72	19.05 ± 34.92	5.34 ± 3.23	5.20 ± 2.84
Range	1.10–123.40	2.00–123.40	2.00–123.40	2.00–12.70	2.00–7.40
Moderate OSA, n (%)	52 (29.9)	7 (4.0)	5 (2.9)	2 (1.1)	2 (1.1)
Severe OSA, n (%)	26 (14.9)	5 (2.9)	4 (2.3)	1 (0.6)	0 (0.0)
O2 nadir, %
Mean ± SD	87.1 ± 7.0	87.5 ± 4.3	87.0 ± 4.8	88.9 ± 2.7	87.3 ± 3.1
Range	50.0–97.0	78.0–94.0	80.0–94.0	84.0–94.0	84.0–90.0
Concurrent surgeries, n (%)
Turbinate reduction	64 (36.8)	8 (4.6)	3 (1.7)	4 (2.3)	1 (0.6)
Adenoidectomy	80 (46.0)	12 (6.9)	5 (2.9)	5 (2.9)	1 (0.6)
Midline posterior glossectomy	35 (20.1)	13 (7.5)	5 (2.9)	7 (4.0)	4 (2.3)
Supraglottoplasty	98 (56.3)	12 (6.9)	5 (2.9)	5 (2.9)	3 (1.7)
No concurrent surgery	13 (7.5)	1 (0.6)	1(0.6)	0 (0.0)	0 (0.0)

AHI = apnea-hypopnea index, ASA = American Society of Anesthesiologists, ED = emergency department, O₂ = oxygen saturation, OSA = obstructive sleep apnea, PSG = polysomnography, SD = standard deviation.

Chi-squared analysis identified a significantly higher proportion of patients with reported complications to have developmental delay when compared to the entire patient population (P = .016). No difference between these two groups was found when isolating all other comorbid conditions. Similarly, Wilcoxon rank sum test revealed no significant difference (P > .05) in age, BMI z-score, mean preoperative AHI, sex assigned at birth, and American Society of Anesthesiologists classification among patients with reported complications compared to those without complications (Table 1).

Only 6 patients (3.4%) were admitted for a duration longer than 24 hours after surgery. Univariate logistic regression modeling did not identify a correlation between prolonged length of stay (> 24 hours) and severe OSA; sex assigned at birth; Trisomy 21; or history of gastroesophageal reflux, asthma, allergic rhinitis, history nonfebrile seizure, or attention deficit hyperactivity disorder or other neuropsychiatric comorbid condition. In contrast, developmental delay was found to be associated with prolonged length of stay (P = .030) with an OR of 11.15 (95% CI: 1.74–216.55). Additional univariate logistic regression modeling did not identify a relationship between age > 8 years old, sex assigned at birth, BMI z-score, American Society of Anesthesiologists class, or history of adenoidectomy or tonsillectomy with any of the postoperative complication metrics studied (Table 4). However, concurrent MPG performed with LT was associated with both increased overall risk of postoperative complication (P = .001) and ED presentation for poor oral intake or pain control (P < .001), with an OR of 5.73 (95% CI: 2.35–14.16) and 11.33 (95% CI: 2.96–55.12), respectively. No other concurrent surgery was associated with risk of complication or failure including turbinate reduction, adenoidectomy, or supraglottoplasty. Additional logistic regression modeling revealed that both asthma (P = .033) and developmental delay (P = .016) were related to having a postoperative complication with an OR of 2.51 (95% CI: 1.08–5.93) and 2.85 (95% CI: 1.22–6.76), respectively. Further analysis demonstrated that asthma was significantly associated with poor oral intake requiring ED visitation (P = .028) with OR of 4.75 (95% CI: 1.27–22.72). No other condition was significantly related to postoperative complications (Table 5).

Table 4.

Clinical characteristics for patients with preoperative and postoperative PSG.

Patient Characteristics	Patients with Pre and Post PSG Data	Surgical Failure	P a
	(n = 145)	(n = 25)
Age (years)			.810
Mean ± SD	8.17 ± 3.54	8.34 ± 3.86
Range	1.89–15.62	2.44–14.78
Age > 8 years, n (%)b	71 (47.9)	12 (8.3)	>.999
BMI z-score			.258
Mean ± SD	0.92 ± 1.43	1.08 ± 1.73
Range	−3.69 to 4.22	−3.69 to 3.04
BMI z-score > 1, n (%)b	76 (51.4)	14 (9.7)	.663
BMI z-score > 2, n (%)b	43 (29.1)	12 (8.3)	.026
Sex assigned at birth, n (%)b			.899
Male	85 (57.4)	15 (10.3)
Female	63 (42.6)	10 (6.9)
Surgical history, n (%)b
Adenoidectomy	99 (66.9)	19 (13.1)	.356
Tonsillectomy	84 (56.8)	17 (11.7)	.270
ASA class, n (%)b			.198
ASA class I	1 (0.7)	0 (0.0)
ASA class II	94 (63.5)	12 (8.3)
ASA class III	53 (35.8)	13 (9.0)
ASA class IV	0 (0.0)	0 (0.0)
Preoperative PSG study
AHI			.005
Mean ± SD	8.06 ± 14.26	20.73 ± 31.11
Range	1.10–123.40	1.10–123.40
Moderate OSA, n (%)b	67 (45.3)	17 (11.7)	.026
Severe OSA, n (%)b	21 (14.2)	10 (6.9)	.002
O2 nadir (%)			.002
Mean ± SD	87.0 ± 6.7	81.5 ± 10.1
Range	52.0–97.0	52.0–92.0
Postoperative PSG study
AHI			<.001
Mean ± SD	3.97 ± 7.74	14.43 ± 14.91
Range	0.00–54.46	5.04–54.46
O2 nadir (%)			<.001
Mean ± SD	89.8 ± 5.9	85.7 ± 6.5
Range	50.0–99.0	66.0–98.0
Comorbid conditionb
Obesityc	56 (37.8)	13 (9.0)	.175
Reflux	53 (35.8)	12 (8.3)	.176
Asthma	56 (37.8)	10 (6.9)	.824
Allergic rhinitis	50 (33.8)	10 (6.9)	.493
ADHD / behavior	45 (30.4)	6 (4.1)	.634
Developmental delay	51 (34.5)	12 (8.3)	.165
Trisomy 21	9 (6.1)	5 (3.4)	.007
Nonfebrile seizure	27 (18.2)	7 (4.8)	.167

^aP values listed correspond to Wilcoxon rank sum test for continuous variables and Fisher’s exact test for categorical variables. ^bThe parentheses in these cells correspond to the percentage of patients out of the 145 patients with preoperative and postoperative PSG data. ^cObesity was defined as patients with BMI z-scores ≥ 1.645 (corresponding to the 95th percentile) in accordance with the American Academy of Pediatrics. ADHD = attention deficit hyperactivity disorder, AHI = apnea-hypopnea index, ASA = American Society of Anesthesiologists, behavior = other behavioral diagnosis, BMI = body mass index, O₂ = oxygen saturation, OSA = obstructive sleep apnea, PSG = polysomnography.

Table 5.

Complication logistic regression models.

Clinical Variable	Any Complication		Postoperative Bleeding		Readmission for Poor Oral Intake or Pain Control		ED Visit for Poor Oral Intake or Pain Control		Surgical Failureb
	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)
Age > 8 years	.849	1.09 (0.47–2.52)	.901	1.09 (0.47–2.52)	.944	1.07 (0.13–9.11)	.911	1.08 (0.29–4.00)	.998	1.00 (0.42–2.38)
Male	.155	1.91 (0.81–4.91)	.189	2.46 (0.71–11.39)	.459	2.34 (0.29– 47.90)	.809	1.17 (0.32–4.74)	.776	1.14 (0.48–2.80)
Female	.155	0.52 (0.20–1.24)	.189	0.41 (0.09–1.42)	.459	0.42 (0.02–3.37)	.809	0.85 (0.21–3.09)	.776	0.88 (0.36–2.10)
ASA class > II	.228	1.68 (0.71–3.91)	.651	1.32 (0.38–4.31)	NAa	NAa	.335	1.88 (0.50–7.01)	.068	2.25 (0.94–5.44)
BMI z–score > 1	.190	1.79 (0.77–4.44)	.135	2.79 (0.80–12.90)	.889	0.87 (0.10–7.37)	.822	0.86 (0.23–3.22)	.610	1.25 (0.53–3.03)
BMI z–score > 2	.417	1.43 (0.59–3.36)	.165	2.31 (0.70–7.72)	.435	2.21 (0.26–18.80)	.910	0.92 (0.19–3.47)	.025	2.74 (1.12–6.68)
History of adenoidectomy	.880	0.94 (0.40–2.33)	.528	0.68 (0.21–2.39)	.722	1.51 (0.19–30.97)	.818	1.18 (0.31–5.63)	.292	1.70 (0.66–4.96)
History of tonsillectomy	.929	1.04 (0.45–2.47)	.917	1.07 (0.33–3.73)	.779	0.75 (0.09–6.39)	.838	1.15 (0.32–4.62)	.217	1.78 (0.73–4.64)
Severe OSA preoperatively	.360	1.67 (0.51–4.72)	.057	3.56 (0.87–12.83)	NAa	NAa	.722	0.68 (0.04–3.95)	<.001	6.61 (2.39–18.46)
Concurrent surgeries
Turbinate reduction	.492	0.73 (0.28–1.74)	.386	0.55 (0.12–1.93)	.625	0.57 (0.03–4.53)	.828	1.16 (0.29–4.21)	.254	1.66 (0.69–3.97)
Adenoidectomy	.984	1.01 (0.43–2.33)	.756	0.829 (0.24–2.70)	.411	0.38 (0.02–3.07)	.793	1.19 (0.32–4.42)	.606	1.25 (0.53–3.00)
Midline posterior glossectomy	.001	5.73 (2.35–14.16)	.065	3.14 (0.88–10.53)	NAa	NAa	<.001	11.33 (2.96–55.12)	.507	0.68 (0.19–1.97)
Supraglottoplasty	.260	0.62 (0.26–1.43)	.295	0.53 (0.15–1.73)	.459	2.37 (0.30–48.41)	.679	0.76 (0.21–2.84)	.148	1.96 (0.81–5.11)
No concurrent surgery	.492	0.73 (0.28–1.74)	.906	0.88 (0.15–16.76)	.991	NAa	.990	NAa	.994	NAa
Comorbid condition
Obesityc	.593	1.26 (0.54–2.91)	.225	2.09 (0.64–7.33)	.725	1.43 (0.17–12.14)	.927	0.94 (0.23–3.42)	.135	1.94 (0.81–4.68)
Reflux	.088	2.08 (0.89–4.88)	.897	0.92 (0.24–3.06)	.132	5.79 (0.72–118.54)	.315	1.93 (0.52–7.20)	.167	1.85 (0.77–4.43)
Asthma	.033	2.51 (1.08–5.93)	.620	1.35 (0.39–4.43)	NAa	NAa	.028	4.75 (1.27–22.72)	.807	1.12 (0.45–2.67)
Allergic rhinitis	.988	1.01 (0.40–2.37)	.195	0.36 (0.05–1.42)	.516	1.93 (0.23–16.43)	.295	1.98 (0.53–7.41)	.472	1.38 (0.56–3.32)
ADHD / behavior	.377	1.48 (0.61–3.47)	.413	1.65 (0.47–5.41)	.095	7 (0.87–143.36)	.530	1.52 (0.38–5.56)	.447	0.68 (0.23–1.75)
Developmental delay	.016	2.85 (1.22–6.76)	.061	3.14 (0.96–11.05)	.110	6.44 (0.80–131.93)	.241	2.15 (0.58–8.06)	.122	1.99 (0.82–4.78)
Trisomy 21	NAa	NAa	NAa	NAa	NAa	NAa	NAa	NAa	.005	7.44 (1.82–32.37)
Nonfebrile seizure	.838	1.12 (0.35–3.04)	.914	0.92 (0.14–3.72)	.121	4.86 (0.57–41.88)	.309	2.08 (0.43–8.00)	.172	2.00 (0.70–5.28)

^aNA appears for OR where there were no observed patients found to have both variables. For example, of the 10 patients diagnosed with Trisomy 21, none of these patients were reported to have a complication. Of note, all patients who required readmission for poor oral intake or poor paint control had a history of asthma and underwent concurrent midline posterior glossectomy. ^bSurgical failure was defined as postoperative apnea-hypopnea index > 5. ^cObesity was defined as patients with BMI z-scores ≥ 1.645 (corresponding to the 95th percentile) in accordance with the American Academy of Pediatrics. ADHD = attention deficit hyperactivity disorder, ASA = American Society of Anesthesiologists, behavior = other behavioral diagnosis, BMI = body mass index, CI = confidence interval, ED = emergency department, NA = not applicable, OR = odds ratio, OSA = obstructive sleep apnea.

Surgical success

Both preoperative and postoperative PSG data were available for 145 patients (83.3%) (Table 3). The mean interval between preoperative PSG and postoperative PSG was 10.2 ± 5.8 months. In this subset of patients, there was significant improvement (P < .001) in mean AHI (preoperative AHI 8.06 ± 14.3, range 1.10–123.40 vs postoperative AHI 4.02 ± 7.81, range 0.00–54.46). Similarly, there was significant improvement (P < .001) in mean oxygen saturation nadir (preoperative oxygen saturation nadir 87.0 ± 6.7%, range 52.0%–97.0% vs postoperative oxygen saturation nadir 89.8 ± 5.9%, range 50.0%–99.0%).

Table 3.

Surgical outcomes.

Measured Outcome	All Patientsa	Patients with Pre and Post PSG Datab	Surgical Failurec	P d
	(n = 174)	(n = 145)	(n = 25)
Preoperative PSG
Preoperative AHI				.285
Mean ± SD	7.88 ± 13.42	8.06 ± 14.3	20.73 ± 31.11
Range	1.10–123.40	1.10–123.40	1.10–123.40
Preoperative O2 nadir				.301
Mean ± SD	87.1 ± 7.0	87.0 ± 6.7	81.5 ± 0.10
Range	50.0–97.0	52.0–97.0	52.0–92.0
Moderate OSA, n (%)	52 (29.9)	48 (33.1)	7 (28.0)	.103
Severe OSA, n (%)	26 (14.9)	21 (14.5)	10 (40.0)	.546
Postoperative PSG
Postoperative AHI	NAe			NAe
Mean ± SD		4.02 ± 7.81	14.91 ± 14.91
Range		0.00–54.46	5.04–54.46
Postoperative O2 nadir	NAe			NAe
Mean ± SD		89.8 ± 5.9	85.7 ± 6.5
Range		50.0–99.0	66.0–98.0
AHI improvement	NAe			NAe
Mean ± SD		4.04 ± 8.70	6.30 ± 19.92
Range		−32.00 to 70.92	−32.00 to 70.92
O2 nadir improvement (%)	NAe			NAe
Mean ± SD		2.7 ± 7.4	4.2 ± 10.3
Range		−35.0 to 36.0	0.0–0.1
Concurrent surgeries, n (%)
Turbinate reduction	64 (36.8)	54 (37.2)	12 (48.0)	.836
Adenoidectomy	80 (46.0)	70 (48.3)	13 (52.0)	.222
Midline posterior glossectomy	35 (20.1)	29 (20.0)	4 (16.0)	>.999
Supraglottoplasty	98 (56.3)	79 (54.5)	17 (68.0)	.310
No concurrent surgeries	13 (7.5)	10 (6.9)	0 (0.0)	>.999
Length of stay				.258
Mean ± SD	1.05 ± 0.34	1.03 ± 0.16	1.08 ± 0.27
Range	1.00–5.00	1.00–2.00	1.00–2.00
Reported complication, n (%)	26 (14.9)	19 (12.8)	4 (16.0)	.153
Postoperative bleeding reported	12 (6.9)	9 (6.1)	2 (8.0)	.424
Self-resolving bleeding	5 (2.9)	4 (2.7)	1 (4.0)	>.999
Required return to the operating room	7 (4.0)	5 (3.4)	1 (4.0)	.330
ED visit for poor oral intake or pain control	10 (5.7)	9 (6.1)	1 (4.0)	.167
Required readmission	4 (2.3)	3 (2.0)	1 (4.0)	.521
Prolonged dysphagiaf	2 (1.1)	1 (0.7)	1 (4.0)	.306
Infection	1 (0.6)	1 (0.7)	1 (4.0)	>.999
ED visit for globus sensation	1 (0.6)	0 (0.0)	0 (0.0)	.167
Required reintubation	1 (0.6)	0 (0.0)	0 (0.0)	.167

^aFor the “All Patients” column, the percentage values are calculated using 174 total patients. ^bFor the “Patients with Pre and Post PSG Data” column, the percentage values are calculated using 145 total patients. ^cFor the “Surgical Failure” column, the percentage values are calculated using 25 total patients. ^dP values listed correspond to Wilcoxon rank sum test for continuous variables and Fisher’s exact test for categorical variables when comparing the “all patients” group to the “patients with pre and post PSG data” group. ^eAHI improvement and oxygen saturation improvement are only available for 145 of the 174 patients. For this reason, these values are marked NA for the “all patients” group. ^fOne patient with prolonged dysphagia was also included in the group of patients corresponding to ED visit for poor oral intake or paint control. This patient was counted only once for the total number of patients with reported complications (n = 26). AHI = apnea-hypopnea index, ED = emergency department, NA = not applicable, O₂ = oxygen saturation, OSA = obstructive sleep apnea, PSG = polysomnography, SD = standard deviation.

Surgical failure, as defined by postoperative AHI ≥ 5, was identified in 25 patients (17.2%). The comorbid conditions for this subset of patients are summarized in Table 3. In the surgical failure group, there was a significantly higher preoperative AHI (P = .005), higher postoperative AHI (P < .001), lower preoperative oxygen saturation nadir (P = .002), and lower postoperative oxygen saturation nadir (P < .001) (n = 25). Logistic regression modeling (Table 5) revealed surgical failure to be associated with BMI z-score > 2 (P = .025), with an OR of 2.74 (95% CI: 1.12–6.68) and Trisomy 21 (P = .005) with an OR of 7.44 (95% CI: 1.82–32.37). Similarly, chi-squared analysis revealed a significant difference in the number of patients with Trisomy 21 (P = .007) and with BMI z-score > 2 (P = .022) when comparing the surgical failure group (n = 25) to all patients with preoperative and postoperative PSG data available (n = 145). Multivariate logistic regression modeling for surgical success (AHI < 5) with concurrent DISE-directed procedures as independent variables—including inferior turbinate reduction, adenoidectomy, MPG, and supraglottoplasty—revealed none of these procedures to have a statistically significant effect on odds of surgical success (Table 6). There were 13 patients who underwent LT alone; 11 of these patients had preoperative and postoperative sleep studies available for comparison. All 11 patients had postoperative AHI < 5 and were considered to be in the surgical success group.

Table 6.

Multivariate logistic regression modeling for surgical success (AHI < 5).

Concurrent Surgeries	P	Odds Ratio (95% CI)
Turbinate reduction	.282	0.584 (0.232–1.465)
Adenoidectomy	.777	0.928 (0.374–2.303)
Midline posterior glossectomy	.687	1.161 (0.380–4.353)
Supraglottoplasty	.143	0.495 (0.188–1.213)

AHI = apnea-hypopnea index, CI = confidence interval.

DISCUSSION

Surgical efficacy

Although LT has previously been shown to be effective in improving AHI, oxygen saturation nadir, and sleep symptoms for children with OSA and obstructive lingual tonsils, there are only a few studies reporting on complications and safety in patients undergoing DISE-directed multilevel sleep surgery.^26,31,32,42 Two meta-analyses of outcomes after LT in children have reported surgical success (AHI < 5) of 51% and 52%.^32,43 Our study demonstrates excellent outcomes with surgical success observed in 120 of the 145 patients (82.8%). Furthermore, this may underrepresent the true surgical success rate of our entire population (174 patients) because some patients, who were symptom free at their postoperative visit, did not undergo a postoperative PSG. These children who did not undergo postoperative PSG were, therefore, unavailable for efficacy analysis. Additionally, a recently published large study of DISE-directed multilevel sleep surgery involving the tongue base for pediatric OSA demonstrated a 25% failure rate for LT and MPG, defined as patients who had either obstructive AHI > 5 or obstructive AHI > 1 with persistent OSA symptoms.³³ Although a lower threshold for surgical failure was used, we found this comparable to the 17.2% of patients in our study with documented AHI > 5 postoperatively.³³

One explanation for the higher surgical success seen in this study as compared to the prior meta-analyses for pediatric LT could be the fact that LT in this study was combined with other DISE-directed interventions. The multivariate logistics regression modeling for surgical success (AHI < 5) with concurrent DISE-directed procedures as independent variables did not reveal any of the concurrent surgeries to be independently correlated with surgical success. Furthermore, although infrequently performed, isolated LT was successful in all 11 patients with preoperative and postoperative PSG available for comparison. These findings suggest that the known efficacy of LT may be further potentiated with concurrent DISE-directed surgical interventions for unique and individual airway obstructive patterns. Thus, combining LT with additional DISE-directed intervention may improve versatility and overall surgical success. This further establishes the importance of understanding the safety of LT in combination with other DISE-directed procedures. Future studies directly comparing isolated LT to DISE-directed multilevel sleep surgery involving LT could further characterize the additive effect of these procedures in cohort.

A previous study demonstrated surgical success (defined by obstructive AHI < 5) of LT in treating persistent OSA after AT in patients with Trisomy 21 to be 61.9%.⁴⁴ In our study, surgical success in patients with Trisomy 21 was 44.4%. The OR of surgical failure for patients with Trisomy 21 in our study was 7.44 (95% CI: 1.82–32.37). These findings reinforce the known difficulty of treating OSA in this patient population. Additionally, overweight children have been shown to have significantly higher postoperative AHI and lower cure rates when compared to healthy weight children.⁴⁵ Our study supported this finding with surgical failure 3 times more likely in children with a BMI z-score > 2.

Postoperative complications

Overall, our study demonstrated LT in conjunction with DISE-directed multilevel surgery to be a safe and effective surgical treatment for a pediatric population. However, surgical complications including postoperative bleeding, poor pain control, and poor oral intake can occur. Of the 132 patients in the most recent meta-analysis of LT as a single-site surgery, adverse events occurred in 14 patients (10.6%).³² This rate is comparable to the 26 cases of 174 patients (14.9%) identified in our study. A previous study surveying 92 children who underwent LT reported bleeding in 4 patients (4.4%), including 2 children (2.2%) who required operative control of bleeding.³⁴ These results are comparable to our findings of postoperative bleeding in 12 patients (6.9%) with 7 (4.0%) requiring operative intervention.

Respiratory complications requiring supplemental oxygen therapy, corticosteroid administration, or intubation have been reported in at least three studies.^30,35,36 In one study of 16 children who underwent LT, 3 developed postoperative airway obstruction and oxygen desaturation related to tongue base edema.³⁵ Although these patients required nasopharyngeal airway placement and supplemental oxygen, reintubation was not needed.³⁵ There has also been 1 patient reported in a previous study who developed pneumonia postoperatively and subsequently required intubation, ventilator support, and a 14-day hospital stay.³⁰ We identified 1 patient (0.6%) who required reintubation immediately following surgery for presumed airway edema and was successfully extubated following 3 days of steroid treatment. No other respiratory complications were identified. Our study did uniquely highlight an increased risk of complications (P = .001) as well as ED presentation (P < .001) when concurrent MPG was performed with LT. This finding may be secondary to inherently more extensive intervention at the tongue base further hindering oral intake and adequate swelling and pain control and is worth further investigation. Overall, our data supports DISE-directed multilevel sleep surgery with LT as a safe and effective treatment option for recurrent or persistent OSA and for patients who are not likely to benefit from primary AT based on DISE.

Study limitations

Our study has several limitations. First, most of the patients in this study (161 patients, 92.5%) underwent additional concurrent airway surgeries at the time of LT to treat several identified obstructive sites in order to limit multiple returns to the operating room and additional anesthetic exposure and inductions. The inclusion of patients who underwent concurrent surgeries at the time of LT limits the ability to directly compare these results with previous single-site LT studies. However, inclusion of these patients allowed for a larger sample size that better captures the possible complications following LT in patients with multilevel sleep surgery. Additionally, all patients in our study underwent LT by coblation despite a variety of different surgical techniques available for LT.^30,36 This may limit the external generalizability of our results to other LT techniques. DISE was used to identify retrolingual obstruction caused by the lingual tonsil in this study. However, other studies have identified lingual tonsil hypertrophy by radiologic imaging modalities such as computed tomography, magnetic resonance imaging, and cine magnetic resonance imaging.^43,46 More research is needed to determine the role of diagnostic imaging in the management of children with OSA.

Finally, it is understood that OSA may improve with observation for a subset of patients due to airway growth as they age. In this study, patients were recommended to complete a postoperative PSG no sooner than 3 months after surgical intervention. This recommendation follows the recently published expert consensus statement for the management of pediatric persistent OSA after adenotonsillectomy.⁴¹ Patients generally underwent preoperative PSG a few months prior to surgical intervention (5.5 ± 3.5 months) allowing for initial consultation in clinic, operative scheduling, and perioperative workup as necessary. Nevertheless, the surgical success observed in this study may have been partially influenced by the effects of airway growth given the observed mean interval between preoperative PSG and postoperative PSG (10.2 ± 5.8 months).

CONCLUSIONS

LT has emerged as an effective option for pediatric OSA patients over the past several years. To our knowledge, our study is the largest to demonstrate the specific postoperative outcomes and complications associated with multilevel sleep surgery involving LT for the treatment of pediatric OSA. Our study demonstrates excellent outcomes (82.8%) and infrequent complications including postoperative bleeding (6.9%) and readmission for poor oral intake or poor pain control (2.3%). Developmental delay was associated with prolonged length of stay while concurrent MPG and history of asthma were associated with increased ED visitation or readmission for poor oral intake. Surgical failure was higher in patients with BMI z-score > 2 and Trisomy 21. LT is a safe and effective surgical procedure for pediatric OSA when performed alone or when performed in combination with other upper airway procedures.

DISCLOSURE STATEMENT

All authors have contributed to, reviewed, and approved the manuscript in its current form. Institution: West Virginia University, Morgantown, West Virginia, USA. The authors report no conflicts of interest.

ABBREVIATIONS

AHI

apnea-hypopnea index

AT

adenotonsillectomy

BMI

body mass index

CI

confidence interval

DISE

drug-induced sleep endoscopy

ED

emergency department

LT

lingual tonsillectomy

MPG

midline posterior glossectomy

OR

odds ratio

OSA

obstructive sleep apnea

PSG

polysomnography