Transoral Robotic Surgery for Pediatric Upper Airway Pathology: An Institutional Update

Abstract

Objective:

Provide an update on our institution’s experience with utilizing transoral robotic surgery (TORS) in pediatric airway surgery and compare these results to surgery by traditional methods.

Methods:

Pediatric patients who underwent TORS for treatment of upper airway pathology between 2010–2021 at our institution were retrospectively identified and compared to patients with the same or similar pathology who underwent a traditional (open or endoscopic) surgical approach over the same time period. Outcomes of interest included patient demographics, operative times, adverse events, hospital length of stay (LOS), and modified barium swallow (MBSS) results.

Results:

Forty children (19M, 21F) underwent 46 TORS procedures. Mean age was 6.4 years (range: 6 days-17 years). Most commonly treated pathology included: laryngeal clefts (LC) (n=18), lymphatic malformations (n=9), and base of tongue masses (n=7). Surgical time was decreased in traditional type I LC repairs (mean: 111 vs 149 min, P = 0.04) and lymphatic malformation excisions (59 vs 120 min, p=0.005). Hospital LOS was increased in TORS type I LC repairs (2.6 vs 1.2 days, P = 0.04). Adverse event rate was similar between TORS and traditional cohorts (17% vs 16% cases, P = 0.9). Postoperative MBSS results were improved for TORS type I LC repairs at 6 months (70% vs 33%, P = 0.09) and 12 months (82% vs 43%, P = 05).

Conclusions:

Pediatric TORS is practical and safe and has comparable outcomes to traditional surgery. Robotic-assisted LC repair displayed improved postoperative swallow results versus traditional approaches and may be particularly useful in recurrent cases.

1. Introduction:

In 2009, da Vinci surgical systems (Intuitive Surgical, Sunnyvale, CA) first received FDA clearance for treating benign and specific malignant tumors of the head and neck [1]. Since that time, numerous authors have investigated the feasibility of utilizing transoral robotic surgery (TORS) for off-labeled uses [2–18]. One such application is in the treatment of upper airway pathology in the pediatric population. As the surgical robot can allow for tremor filtration, bimanual instrumentation, and improved three-dimensional visualization of critical structures, the applications of TORS in pediatric airway surgery appears sensible.

In 2016, our institution published one of the largest case series of that time [19], which demonstrated the safety and efficacy of utilizing TORS in the treatment of pediatric upper airway pathology in appropriately-selected patients. In this prior study, sixteen children underwent eighteen TORS procedures in the treatment of a wide variety of pathology including: type I, II, III laryngeal clefts, aerodigestive tract strictures, lymphatic malformations, saccular cysts, and base of tongue masses. Since the time of our original publication, we have more than doubled the number of TORS cases performed at our institution in addition to expanding our robotic indications. However, minimal objective data exists in the current literature comparing surgical outcomes between TORS and traditional approaches in the treatment of pediatric upper airway pathology. Thus, the objective of this report is two-fold: 1) provide an update on our institution’s experience with utilizing TORS in pediatric airway surgery and lessons learned and 2) compare surgical outcomes between robotic and traditional approaches to upper airway pathway in an effort to evaluate its effectiveness. From this, we hypothesize that TORS approaches will display improved postoperative surgical outcomes with similar operative time metrics, adverse event rates, and hospital length of stay compared to traditional approaches.

2. Materials and Methods

This study received approval from the University of North Carolina at Chapel Hill Institutional Review Board. Pediatric patients who underwent TORS for treatment of upper airway pathology at the University of North Carolina Hospitals between January 2010-March 2021 were identified via retrospective review of the electronic medical record. Pediatric patients who underwent a traditional approach (endoscopic or open) for treatment of either the same or similar pathology over the same time period (January 2010-March 2021) were also identified and used as matched controls for comparison. For both groups, basic patient characteristics including age, sex, weight, and American Society of Anesthesiologists (ASA) score from the day of surgery were recorded. For patients undergoing laryngeal cleft repair, the presence of any syndrome or neuro-developmental comorbidity that could potentially impact oropharyngeal dysphagia were documented.

Selection criteria for a traditional versus robotic approach was multifactorial including the patient’s size at time of surgery, presence of airway compromise, severity of aspiration or pulmonary complications, practice pattern changes, surgeon and caregiver preference, and hospital robot availability. In general, any patient less than 5 kg was not considered for robotic surgical repair. There are surgeons at our institution that are TORS-certified and offer this approach to patients. For those surgeons that are institutionally TORS-certified, both TORS and traditional approaches are offered to the patient and caregivers. Caregivers were fully informed on the risks, benefits, and alternatives for the different surgical approaches and about the off-labeled nature of using the surgical robot in children. Ultimately, caregiver’s preferences were always respected and a shared decision was made between the provider and caregiver on the best surgical approach for the patient.

Operative time calculations were performed in a similar manner to prior [19]. Briefly, intraoperative time stamp data were reviewed and the following outcomes were calculated: 1) operative set-up time (duration from release of patient from anesthesia team to pre-incision time out), 2) surgical time (duration of surgical procedure), and 3) operating room (OR) time (duration from patient entering to leaving OR, excluding any PACU holds). Next, any intraoperative complications, postoperative adverse events, and hospital length of stay (LOS) were documented. In this study, a postoperative adverse event was defined as any unplanned reintubation in the postoperative hospital course or surgical complication necessitating re-operation. Finally, for patients who underwent laryngeal cleft repair, postoperative swallow function improvement was determined by comparing each patient’s pre- and post-operative modified barium swallow study (MBSS). Improvement in MBSS was characterized by either: 1) a transition from preoperative MBSS revealing laryngeal aspiration to postoperative MBSS displaying laryngeal penetration or a normal swallow, or 2) a shift from preoperative MBSS indicating laryngeal penetration to postoperative MBSS showing a normal swallow. For patients with multiple postoperative MBSS results, improvement was determined by comparing the patient’s best MBSS result at 6-, 12-, and 24-months following surgery to their preoperative result. Both speech-language pathology and radiology were blinded to the surgical approach used for each patient when conducting and interpreting the MBSS.

2.1. Statistical Analysis

Due to limited sample sizes, statistical analyses were not performed between TORS and traditional groups for type III laryngeal cleft repairs, laryngotracheal reconstructions (LTR) for posterior glottis stenosis or bilateral vocal fold immobility, or for saccular cyst excisions. For the remaining pathology groups, chi-square (χ2) analysis was used to compare dichotomous variables between TORS and traditional cohorts, while Welch’s t test was used to compare continuous variables. All statistical analyses were performed utilizing GraphPad Prism, software package version 9.5.1 (GraphPad Software, Inc).

3. Results

Forty children (19M, 21F) underwent 46 TORS procedures (Table 1). Of this total, sixteen patients and eighteen procedures were included from our prior study [19]. For new TORS cases since 2016, the surgical robot and operating room were setup in the same manner as prior (Figure 1) [11]. Over the same time period, 45 children (22M, 23F) underwent 57 procedures utilizing traditional approaches for treatment of same or similar pathology (Table 1). The mean age in the TORS group was 6.4 years (range: 6 days-17 years), and in the traditional group was 5.7 years (range: 5 days-18 years). Mean weight in the TORS group was 23.3 kg (range 2.4–93.7), and in the traditional group was 22.1 kg (range 4.2–71.5). In general, there were few statistically significant differences in patient characteristics between TORS and traditional groups when compared by pathology (Table 1). The main exception to this was that ASA score was higher in robotic vs traditional Type I laryngeal cleft repairs (mean: 3.1 vs 2.4, P = 0.008), while ASA score was lower in robotic vs traditional BOT mass excisions (mean: 2.0 vs 3.0, P = 0.04) (Table 1). In the TORS vs traditional cohorts, syndromic or neuro-developmental comorbidities potentially contributing to oropharyngeal dysphagia were seen in 55% vs 58% of type I laryngeal cleft patients, 33% vs 67% of type II laryngeal cleft patients, and 67% vs 50% of type III laryngeal cleft patients, respectively.

Table 1.

Patient Characteristics in TORS and Traditional Group by Pathology

		Pathology
	Approach	Type I LC	Type II LC	Type III LCa	Lymphatic Malformation	BOT Mass	PGS/BVFIa	Aerodigestive Tract Stricture	Saccular Cyst/Neurofibromaa
Procedures (number)	TORS	11	4	3	9	7	2	6	4
	Traditional	19	9	2	8	6	6	5	2
Age (mean, years)	TORS	4.0	1.5	3.9	11.6	7.9	9.3	8.7*	41 days
	Traditional	3.3	2.6	46 days	10.7	9.1	4.6	13.2*	3.9
Sex (M:F)	TORS	7:4	3:0	3:0	2:5	1:6	0:2	3:1	3:1
	Traditional	12:7	2:4	2:0	1:3	4:2	3:3	1:2	0:1
Weight (mean, kg)	TORS	16.7	11.2	14.6	43.1	26.0	33.0	24.9	3.3
	Traditional	14.4	11.2	4.3	36.8	44.4	18.8	31.6	19.7
ASA score (mean)	TORS	3.1***	3.3	2.7	2.8	2.0*	3.0	2.8	3.5
	Traditional	2.4***	2.9	3.5	3.2	3.0*	2.8	3.4	3.5
Hospital LOS (mean, days)	TORS	2.6*	4.0	4.3	2.6	1.7	4.0	6.0	19.5
	Traditional	1.2*	12.2	94.5	1.6	1.7	15.5	0.2	5.5

LC, Laryngeal cleft. BOT, Base of tongue. PGS, Posterior glottic stenosis. BVFI, Bilateral vocal fold immobility. TORS, Transoral robotic surgery. ASA, American Society of Anesthesiologists. LOS, Length of stay.

^aStatistical analysis not performed due to limited sample size

^*Welch t test p value < 0.05 for difference in means between TORS and traditional groups.

^***Welch t test p value < 0.001 for difference in means between TORS and traditional groups

Figure 1.

Positioning of the patient and surgical robot.

All TORS cases had successful robot access and none required intraoperative conversion to a traditional approach. There were no intraoperative complications noted in the TORS cohort. In the traditional cohort, there was one intraoperative complication of an aspiration event during a single-stage laryngotracheal reconstruction (LTR) procedure in which the patient developed acute onset bradycardia and prolonged desaturations necessitating a brief episode of chest compressions and urgent reintubation. In total, there were eight (17%) postoperative adverse events in the TORS group versus nine (16%) in the traditional group (Table 2). This difference was not statistically significant (P = 0.9).

Table 2.

Postoperative Adverse Events in TORS and Traditional Cohorts

Postoperative Adverse Event	TORSa (n; %)	Traditionalb (n; %)
Surgical failure/Required revision surgery	3 (7%)	9 (16%)
Granulation tissue/Granuloma	2 (4%)	0
Post-extubation	1 (2%)	0
laryngospasm/bronchospasm
Aerodigestive tract stricture	1 (2%)	0
Pneumonia	1 (2%)	0

^a.N = 46 total procedures in TORS cohort

^b.N = 57 total procedures in traditional cohort

In type I laryngeal cleft repairs, both surgical time (mean: 111 vs 149 min, P = 0.04) and OR time (mean: 139 vs 207 min, P = 0.001) were decreased for traditional compared to TORS approaches (Figures 2, 3). Additionally, for lymphatic malformation excisions, both surgical time (mean: 59 vs 120 min, P = 0.005) and OR time (mean: 91 vs 175 min, P = 0.0003) were decreased for traditional compared to TORS approaches (Figures 2, 3). There were no statistically significant differences in surgical time or OR time between TORS and traditional groups for any other pathology. Differences in operative set-up time between TORS and traditional subjects were generally not significant other than lymphatic malformation excisions in which set-up time was greater in TORS compared to traditional approaches (mean: 22 vs 5 min, P = 0.02) (Figure 4).

Figure 2. Comparison of surgical time by pathology between TORS and traditional groups.

Individual data points are presented along with the group mean +/− standard deviation

* p < 0.05

** p < 0.01

Figure 3. Comparison of operating room (OR) time by pathology between TORS and traditional groups.

Individual data points are presented with the group mean +/− standard deviation.

** p < 0.01

*** p < 0.001

Figure 4. Comparison of operative set-up time by pathology between TORS and traditional groups.

Individual data points are presented along with the group mean +/− standard deviation.

* p < 0.05

Postoperative hospital LOS was increased for patients who underwent TORS vs traditional type I laryngeal cleft repair (2.6 vs 1.2 days, P = 0.04). No statistically significant differences in postoperative hospital LOS were observed for any other pathology (Table 1). For type I laryngeal cleft repairs, TORS approaches demonstrated improved postoperative MBSS results compared to traditional approaches at 6 months (70% vs 33%, P = 0.09), 12 months (82% vs 43%, P = 0.05), and 24 months (82% vs 43%, P = 0.05) following surgery (Figure 5). Although sample size was not large enough to allow for statistical analysis, similar trends were seen for type II and III laryngeal cleft repairs, in which TORS approaches demonstrated improved postoperative MBSS results at 6 months (75% vs 22%; 67% vs 50%, respectively) compared to traditional approaches.

Figure 5.

Comparison of modified barium swallow study (MBSS) results and improvement rate between TORS and traditional type I laryngeal cleft repairs at A) 6-, B) 12-, and C) 24-months postoperatively.

* p < 0.05

4. Discussion

Since first described by Rahbar et al [12] in the repair of laryngeal clefts, TORS has been increasingly used in the treatment of pediatric airway pathology. In this study, we show the successful utilization of TORS in the treatment of a wide variety of pediatric upper airway pathology including: type I, II, III laryngeal clefts, lymphatic malformations, lingual tonsil hypertrophy, base of tongue hamartomas, thyroglossal duct cysts, aerodigestive tract strictures, saccular cysts, and bilateral vocal fold immobility. Importantly, we were able to utilize TORS safely and effectively across a wide spectrum of pediatric patients, ranging from 6 days-17 years old and from 2.4–93.7 kg. This highlights the idea that the surgical robot can be safely used even in the smallest neonates.

Historically, some of the predominant concerns with TORS have included the initial high startup cost of purchasing the surgical robot, loss of tactile feedback when operating, and increased OR time required for setting up and docking the robot. As shown above, operative set-up time, surgical time, and OR time were generally comparable between TORS and traditional approaches for most treated upper airway pathology at our institution (Figures 2, 3, 4). The two exceptions being in cases of less complex pathology such as in type I laryngeal clefts and focal lymphatic malformations, in which traditional approaches displayed faster surgical and OR times compared to TORS. In examining our other outcomes of interest, we saw that traditional type I laryngeal cleft repairs had a decreased postoperative hospital LOS compared to robotic repairs (Table 1). This finding could be explained by our TORS cohort having a higher mean ASA score compared to the traditional cohort and thus a higher medically complexity which could necessitate additional hospital resources and time to properly manage. However, interestingly, when analyzing our postoperative swallow outcomes, we saw that patients who underwent TORS type I laryngeal cleft repairs had a higher likelihood of improvement in their postoperative MBSS results at 6 months, 12 months, and 24 months following surgery compared to patients who underwent traditional approaches despite higher ASA score (Figure 5). This data could suggest that robotic assistance offers an improvement in surgical cleft closure technique at our institution. While we recognize that MBSS improvement rates in our traditional endoscopic type I laryngeal cleft repair cohort were lower than the 68% MBSS improvement rate reported in a similar study [20], results can be heavily influenced by the percentage of patients in the cohort who have concomitant neurodevelopmental or swallowing comorbidities thus limiting accurate comparisons. Considering this, a minor increase in surgical time and hospital LOS may be a reasonable trade-off for the significant improvement in postoperative swallow function seen in these patients.

With increasing complex pathology such as in type II and III laryngeal cleft repairs, surgical time and OR time trended towards being decreased for TORS approaches compared to traditional endoscopic or open approaches (Figures 2, 3), although small sample sizes limited statistical comparisons. This observation aligns with the qualitative experience of the senior author in that the three-dimensional visualization and bimanual instrumentation afforded by the surgical robot during cases of complex airway pathology allowed for improved mechanical dexterity, tissue retraction, and more precise suture placement which generally resulted in a more efficient operation. Additionally, postoperative hospital LOS trended towards being decreased for TORS type II and III laryngeal cleft repairs vs traditional approaches (Table I). Similarly, to type I clefts, TORS approaches to type II and III laryngeal cleft repairs also displayed improved postoperative MBSS results at 6 months, 12 months, and 24 months compared to traditional approaches. However again, again, small sample sizes precluded more robust statistical comparisons.

When analyzing our institutional data, we also compared surgical time and OR time over the duration of this study, evaluating if we had improved efficiency when using the surgical robot over time. When we compared our original sixteen TORS cases reported in Zdanski et al. 2017 [19] to the thirty additional cases over the years that have followed, we found that surgical time (mean: 154 vs. 145 min; respectively, P = 0.6) and OR time (mean: 208 vs. 191 min; respectively, P = 0.3) were slightly decreased in our more recent cohort of patients. However, these results were not statistically significant. In analyzing our results in the context of the previously reported literature, our operative time findings for robotic approaches to type I and II laryngeal clefts and aerodigestive tract strictures appear to be generally comparable with prior studies [10,11,15], while our operative times for TORS lingual tonsillectomy appear to be slightly increased compared to a prior case series [3]. Unfortunately for other treated pathology included in this study, there is very limited robotic operative time data in the literature, outside of isolated case reports, available for comparison.

Overall postoperative adverse event rate was similar between TORS and traditional cohorts (17% vs 16% of cases; respectively, P = 0.9). Postoperative adverse events in the TORS group are summarized in Table 2 and consisted of: 1) Patient who was re-intubated in the OR following type I laryngeal cleft repair due to bronchospasm upon extubation, 2) Patient who developed supraglottic and subglottic granulation tissue one month after prolonged intubation for respiratory failure following type I laryngeal cleft repair, 3) Patient with a residual cleft following type II laryngeal cleft repair which was corrected following revision robotic surgery, 4) Patient with a cervicofacial lymphangioma who developed oropharyngeal stenosis following robotic excision who required additional robotic and endoscopic scar release procedures, 5) Two patients with failed robotic LTR who had resorbed posterior cartilage grafts requiring surgical revision (one with a revision endoscopic posterior graft LTR followed by cordotomy; one treated at an outside hospital), 6) Patient who developed pneumonia and septic shock following robotic approach to multilevel aerodigestive tract stricture release requiring emergent bronchoscopy, and 7) Patient with a postoperative granuloma at site of robotic saccular cyst excision requiring laser excision. Postoperative complications in the traditional cohort are summarized in Table 2 and consisted of: 1) Patient with a residual cleft following two traditional type II laryngeal cleft repairs who was successfully revised with a robotic approach, 2) Patient with a residual cleft following three type II laryngeal cleft repairs who was successfully revised with a robotic approach, 3) Patient with a residual cleft following type III laryngeal cleft repair who underwent revision surgery at an outside hospital, 4) Patient with a failed endoscopic posterior cricoid split with posterior graft placement who required cordotomy for persistent stridor, 5) Patient with a failed endoscopic posterior cricoid split with posterior graft placement who underwent surgical revision at an outside hospital, 6) Patient with an incomplete supraglottic neurofibroma resection requiring revision surgery.

A final point of discussion is how TORS was able to be used successfully in patients that had failed prior traditional approaches. In this study, eight (14%) procedures in the traditional group and three (6.5%) procedures in the TORS group required revision surgery. While not statistically significant (P = 0.33), revision rate trended towards being decreased in the TORS group. In the traditional group, five of these eight cases represented two patients: one patient with a residual type II laryngeal cleft after two failed endoscopic repairs, and the other patient with a residual type II laryngeal cleft after two failed open repairs and another failed endoscopic repair. Importantly, residual clefts in these two patients were able to successfully revised with a single TORS approach without further complications. This could suggest the utility of TORS as an alternative surgical approach in cases of recurrent clefts that have failed prior traditional methods of repair.

There are some limitations with this study. Firstly, while every attempt was made to match TORS cases with associated controls with the same pathology, this was not possible in every case. Namely, there were no endoscopic saccular cyst excisions for comparison to robotic approaches, and thus two cases of supraglottic neurofibromas that were endoscopically resected in a similar location and fashion were selected for comparison. We recognize that differences in pathology and age ranges of these patients make comparisons between these groups suboptimal. An additional concern in this study was that the heterogeneity of treated pathology and associated limited sample sizes of the robotic approaches for these cases made comparisons of outcomes difficult. It is our hope that as we continue to expand the numbers of robotic pediatric airway surgeries performed at our institution that future comparisons will be more robust. A final point of discussion is in regards to the surgical robot itself. Over the course of the study we switched to utilizing the da Vinci SP platform from the da Vinci Si platform, which could have introduced potential biases in our analyses. Overall, we feel as though this switch has been a net neutral so far. Negatives of the SP platform such as slower robot set up and docking time and issues with instrument range of motion have been offset by smaller instrument arms that can more effectively delivery energy to more precise locations. Currently, we have yet to conclusively determine if the SP platform is better suited to work in confined spaces such as the oral cavity in neonates and toddlers.

Over our 10 years of institutional experience with pediatric robotic surgery we have learned some meaningful lessons. These include: 1) the importance of a two-surgeon team in maximizing surgical field exposure and assisting the robotic surgeon while protecting the patient particularly with regard to the airway, 2) ensuring that all robotic arms are freely mobile within the surgical field and not at risk of colliding with the patient or one another; 3) appropriate selection of surgical candidates for TORS versus traditional approaches. As the surgical robot continues to improve technologically, we expect that the applications of robotic-assisted surgery in the treatment of pediatric airway pathology will continue to expand.

5. Conclusions

This study represents the largest pediatric TORS case series to date and one of the first to compare outcomes between robotic and traditional approaches in the surgical treatment of pediatric upper airway pathology. In this study, we show that utilizing TORS in the appropriately selected pediatric patient is practical and safe and has comparable operative times, complication rates, and hospital LOS to surgery by traditional methods, except for type I laryngeal clefts and lymphatic malformations in which traditional approaches appear to be faster. Importantly, we found that postoperative swallow outcomes were significantly improved in TORS type I laryngeal cleft repairs and trended towards improvement for TORS type II and III cleft repairs. These findings could suggest that robotic assistance offers an improvement in surgical cleft closure technique at our institution. Finally, TORS may be a particularly useful alternative surgical option in difficult cases that have failed prior traditional approaches.