Authors: Pooja Santapuram (1.Department of Anesthesiology, Columbia University Irving Medical Center), William S. Tierney (2.Cleveland Clinic Voice Center), Li-Ching Huang (3.Vanderbilt University Medical Center, Department of Biostatistics, Nashville, TN), Sheau-Chiann Chen (3.Vanderbilt University Medical Center, Department of Biostatistics, Nashville, TN), Lynn D. Berry (3.Vanderbilt University Medical Center, Department of Biostatistics, Nashville, TN), David O. Francis (4.University of Wisconsin School of Medicine and Public Health, Department of Surgery, Division of Otolaryngology-Head and Neck Surgery, Madison, WI), Alexander Gelbard (5.Department of Otolaryngology-Head and Neck Surgery, Vanderbilt University Medical Center)
Categories: Article, idiopathic subglottic stenosis, endoscopic dilation, comparative techniques
Source: The Laryngoscope
Doi: 10.1002/lary.31393
Authors: Pooja Santapuram, William S. Tierney, Li-Ching Huang, Sheau-Chiann Chen, Lynn D. Berry, David O. Francis, Alexander Gelbard
A small number of Idiopathic subglottic stenosis (iSGS) patients are treated at institutions across the country. Divergence in operative techniques for endoscopic dilation (ED) of iSGS has been anecdotally recognized but not formally characterized. Additionally, the relationship between procedural variation and clinical outcome has not been studied.
Secondary analysis of the NoAAC iSGS^1000^ cohort investigated variation in procedural techniques and treatment outcomes in patients treated with ED across high-enrolling treatment centers (enrolled >10 patients in PR-02 trial).
13 NoAAC centers each enrolled >10 patients treated with ED for a total of 281 subjects. There was significant variation in procedural details and rate of recurrence among institutions. Hierarchal cluster analysis revealed significant heterogeneity among institutions and clusters in all procedural variables. However, analysis demonstrated a transient delay in disease recurrence in cluster 2 which disappeared with longer longitudinal follow-up. Patient-reported outcome and peak expiratory flow data supported the potential benefit of the technical variation in Cluster 2. Distinct to cluster 2, however, was routine use of adjuvant triple medical therapy (proton pump inhibitor (PPI), antibacterial, and steroid inhaler).
Both outcome and procedural technique vary among centers employing ED to treat iSGS. A transient delay in recurrence was observed among centers that routinely prescribed adjuvant medical therapy (antibiotic, inhaled corticosteroid, and PPI) to iSGS patients after endoscopic dilation, which was further supported by patient-reported data and peak expiratory flow data. Prospective studies are needed to understand the effects of adjuvant medical therapy on recurrence after endoscopic dilation.
Idiopathic subglottic stenosis (iSGS) is a rare disease (incidence 1 in 400,000) affecting primarily Caucasian females in adulthood ^1^. A recent pragmatic clinical trial enrolling 810 iSGS patients demonstrated endoscopic dilation (ED) was the most widely employed treatment (nearly 75% of the enrolled patients) ^2,3^. Yet despite therapy, mucosal fibrosis in the proximal airway often recurs, requiring repeated surgeries to maintain airway patency. 50% of iSGS patients treated with ED will recur within 5 years ^3^. Variation in clinical outcome after ED preformed at different institutions has been previously reported ^1^. In an effort to understand this variability, studies have investigated the relationship of patient characteristics ^1^, social determinants of health ^4^, and procedural volume to clinical outcome ^1^. These efforts have not uncovered variables associated with ED treatment response.
The archetypical endoscopic dilation procedure involves endoscopic exposure of the area of stenosis and mechanical dilation of the affected mucosa with rigid instruments or balloon dilators. Many providers also employ “relaxing” mucosal incisions (either sharply or with a laser), inject anti-inflammatory reagents or apply anti-neoplastic medication to the region of stenosis at the time of dilation. Variability in the order of these procedural elements, the instruments employed, and the pharmacological agents administered is substantial. For example, some centers in the NoAAC PR-02 trial used graduated rigid bronchoscopes while the majority expose the stenosis using a laryngoscope followed by balloon dilation to increase airway caliber. Most centers perform intralesional corticosteroid injection at time of dilation, but the specific steroid used and whether the steroid is injected before or after dilation is not standardized. Similarly, anesthesia technique, perioperative medication administration, and post-operative pharmacotherapeutic protocols are divergent. Recent publications investigating adjuvant medical therapies suggested efficacy in slowing the rates of disease recurrence but are confounded by the lack of generalizability in the face of marked variance in procedures between sites.
While variability in procedural technique for ED has been anecdotally recognized, it has never been formally studied. Additionally, the relationship between procedural variation and clinical outcome has not been explored. This study aimed to (1) characterize procedural variability in ED preformed at high-enrolling airway centers, (2) evaluate difference in outcome after ED at these centers, and (3) assess whether ED procedural variations were associated with differential treatment response. We used a mixed-methods approach to address this complex relationship that involved semi-structured interviews to characterize center-level procedural techniques and quantitative methods to evaluate the relationship between procedure variations and treatment outcomes.
The methods describing the consent, study procedures, and data acquisition from the North American Airway Collaborative (NoAAC) iSGS^1000^ cohort in the PR-02 study have been described previously ^1,2,5^.
The prospectively collected NoAAC dataset included adult patients ≥18 years with iSGS meeting established diagnostic criteria. Recruitment of patients took place from June 1, 2015, to June 1, 2017. Additional exclusion criteria have been previously described.^4^ Longitudinal patient data were collected via self-report and automated electronic data capture. Following enrollment, patients completed an electronic health status check every 3 months as well as patient-reported outcome measures tracking symptoms and quality of life every 6 months. Additional data collected included adjuvant medication use, adjuvant medical treatments, and treatment-associated complications. Specific data collection protocols have been previously described ^5^.
To evaluate the variability and effectiveness of endoscopic dilation procedures, data from institutions who enrolled more than 10 subjects managed with endoscopic dilation were included (n=13). Subjects treated at centers reporting 10 or fewer ED-managed patients were excluded from analysis. Patients that underwent endoscopic resection with adjuvant medical therapy (ERMT) or more invasive treatment techniques (e.g., Maddern procedure, laryngotracheoplasty, or cricotracheal resection) were also excluded from analysis.
The primary endpoint was need for and time to recurrent operation (TTR) defined as days from index procedure (T0) to recurrent operation (TR). This endpoint has previously been employed as a valid surrogate for disease recurrence ^2,6^. Additional established secondary endpoints included patient reported outcomes inclusive of voice, and breathing, as well as longitudinal peak expiratory flow rate ^7^.
Semi-structured qualitative assessment of endoscopic dilation procedural details was obtained from 13 high-enrolling centers via direct interview. Interviews were conducted by study investigators (WT, AG) who asked questions verbatim from a survey instrument generated a priori (Supplemental Fig. 1). The surgeon contact for the NoAAC at each site was interviewed. They were asked to comment on the behaviors of all surgeons at the site. Prespecified data regarding anesthetic management, specific dilation technique, and adjuvant medication usage during the peri-operative and post-operative period were collected from each center (Supplemental Table 1). Most sites with multiple surgeons had similar protocols, but when multiple protocols existed this were noted in the survey. The data objects collected were defined via expert consensus of the NoAAC steering committee membership.
Pareto analysis was employed as previously described ^8^. In brief, Pareto analysis is a statistical technique based on the principle of factor sparsity (i.e., roughly 80% of effects are attributable to 20% of causes) and produces a power law probability distribution that graphically displays both the probability density and cumulative distribution. Kruskal-Wallis and Pearson’s Chi-squared tests were used to compare sociodemographic data among centers.
For treatment outcome after endoscopic dilation, the Kaplan-Meier method assessed time to recurrent surgical procedure, and hazard ratios (HRs) and 95% CIs were calculated using Cox proportional hazards regression models. Censoring occurred with either last date of follow-up or death. Fixpoint tests were used to compare the survival curves at a fixed point in time.
For peak expiratory flow (PEF) rate, longitudinal linear regression was used to compare PEF among treatment groups. To account for potential correlation within the same individual, generalized estimating equations were used with the Huber-White method to estimate robust covariance matrix.
To interrogate variation in procedural details among centers, surgeon-reported data underwent dimensionality reduction using hierarchical clustering. Hierarchical cluster analysis was performed using Ward’s method on squared Euclidian distances. The optimal number of clusters was selected by the elbow method. The hierarchical structure of the data was visualized using a dendrogram plot and a cluster plot. Multiple imputation using chained equation was used to generate 1000 values, averaged across imputations to impute the missing score.
Two-sided P values less than or equal to 0.01 were considered statistically significant to adjust for multiple comparisons. Analysis was performed using R version 4.1 and GraphPad Prism 8.3 (GraphPad Software, La Jolla, CA).
Thirteen centers met inclusion criteria and treated 281 patients with endoscopic dilation. Patients treated at these centers were representative of the larger 810 iSGS patient cohort (i.e. 98% female, 97% white, and median age 53 years [IQR: 46 – 61]) (Fig 1A).
Variation in procedural details for ED existed across centers. Twelve (92%) centers used perioperative systemic steroids and 2 (15%) used perioperative antibiotics. All 13 (100%) centers used total intravenous anesthesia (TIVA) with propofol and 11 (85%) used muscle paralysis. For airway management, 10 (77%) centers used supraglottic jet ventilation and 6 (46%) used intermittent apnea.
Surgery-specific technical details also varied by center. Twelve centers (92%) used relaxing incisions; five centers (38%) used a sickle knife to make incisions while 9 (69%) used CO2 laser (one center used both). Twelve centers used balloon dilation alone and one center (8%) used both balloon and Jackson rigid dilator. Of the six centers that reported a balloon pressure, a mean balloon pressure of 6.2 atmospheres was used. No center applied pressure until oxygen desaturation was observed; rather, at 6 centers (46%), pressure was applied for a pre-specified time interval (90 seconds), and in 8 (62%), the dilation process was repeated (for a total of 2 dilations).
All centers used intraoperative intralesional corticosteroid (CS) injections with twelve centers (92%) using triamcinolone and one (8%) using methylprednisolone. CS were used before dilation at 7 centers (58%) and after dilation at 5 (42%) institutions (1 center did not report on this measure). Six centers (46%) used Mitomycin C (MMC; at 0.4 mg/mL for 2 minutes) with 2 centers using it before and 4 after dilation.
For post-operative care, 11 centers (85%) discharged patients the same day, while 2 (15%) admitted patients for 23 hours of monitoring. For post-operative medications, 3 centers routinely prescribed antibiotics (23%), 6 (46%) prescribed inhaled CS, 5 (38%) prescribed proton pump inhibitors (PPIs), and one (8%) prescribed oral steroids. Post-operative adjuvant serial intralesional steroid injections (SILSI) were planned at four centers (31%), utilized for salvage following clinical disease progression at two (15%), and not performed at 7 centers (54%).
To interrogate how ED procedural details impact treatment outcome, we next clustered the 13 centers based on similarities in the technical details employed during ED. Principle component analysis of procedural details yielded three components or center clusters (Fig 2A, 2B, 2C) and dendogram analysis identified the distribution of individual centers within each cluster (as well as quantified the degree of difference between both centers and clusters). Patient characteristics differed between clusters and the specifics of each cluster are listed in Table 1. Cluster 2 had fewer patients (n = 37) than either cluster 1 or 3 (n = 116, 128). Patients in Cluster 3 were 100% white (vs 94% in Cluster 1 & 2, p = 0.32), and 57% of patients in Cluster 2 were peri/postmenopausal (vs. 30% in Cluster 1 and 20% in Cluster 3; p = 0.03). In regard to procedural differences, both sites in Cluster 2 prescribed three post-operative adjuvant antibiotics, inhaled steroids, and proton pump inhibitors. Additionally, both sites in Cluster 2 used intralesional corticosteroid before dilation.
Two-year recurrence rates (defined as repeated operative intervention) following endoscopic dilation varied between 55% and 10% among the 13 centers (p = 0.001) (Fig 1B). The number of patients enrolled at a center was not associated with the observed recurrence rate (p = 0.796) (Fig 1C).
Next, we interrogated clinical outcomes for the three distinct treatment clusters to explore the relationship between procedural variation and clinical outcome. Kaplan Meier curves showcased that Cluster 2 had fewer recurrences initially than Cluster 1 (p = 0.002 at 1.5 years) and Cluster 3 (p = 0.002 at 1.5 years), with longer time until recurrence within the first 2 years following endoscopic dilation (Fig 3A). However, these differences disappeared with a longer period of observation.
Voice Handicap Index (VHI-10) scores did not differ between clusters at any time point (Fig 3B). However, subjective breathing scores (assessed with the COPD Dyspnea Index: CCQ) showed lower scores (less dyspnea) in Cluster 2 at baseline and at 180-day follow up (p = 0.007) (Fig 3C) suggesting less symptomatic dyspnea both initially and at 6-month follow-up. Peak expiratory flow rate (PEFR) in the 3 clusters was also significantly different over time (p = 0.03). Notably, PEFR was higher in Cluster 2 compared to Cluster 3 after 1.5 years (p < 0.03) (Fig 3D).
Studies have shown differences in clinical outcome both within and among disparate surgical approaches to iSGS treatment ^2^. In an effort to understand this variability specifically after endoscopic dilation, prior studies have investigated the relationship of patient characteristics ^1^, social determinants of health ^4^, and procedural volume to clinical outcome ^1^. These efforts did not uncover variables associated with ED treatment response. Now in this work, we explore the relationship between ED technical variation and clinical outcome. We demonstrated variability in disease recurrence after ED in 13 high-enrolling institutions. We also characterized significant ED technical variation between centers in addition to broader practice variation and explored how technical aspects of ED as well as the use of adjuvant therapies influenced treatment outcome.
Cluster analysis allowed exploration of heterogeneity in treatment effect and grouped centers into 3 unique treatment clusters. While none of the subgroups were associated with a significant difference in time-to-reoperation, cluster 2 had fewer early disease recurrences. This may be related to the post-operative triple therapy with antibiotics (trimethoprim/sulfamethoxazole [TMP-SMX] or macrolides), inhaled steroids, and proton pump inhibitors (PPIs). In addition, pre-dilation intralesional corticosteroids may offer a benefit in time to recurrence. However, this effect could result from censoring bias (due to the small number of patients in cluster 2) or the higher percentage of perimenopausal patients in this group. Additionally, baseline CCQ scores were lower in cluster 2, suggesting they may have less severe disease at baseline. This could also account for the transient delay in disease recurrence.
This study adds to the literature on iSGS management documenting the variation in both technique and outcomes after endoscopic dilation. These findings are important and highlight the challenges and limitations of multicenter surgical research. Additionally, while not directly related to iSGS treatment and outcomes, we feel that the methods used in this paper will provide value to investigators exploring alternate multidimensional clinical datasets. Cluster analysis is often used in qualitative research and perceptual research but rarely applied to surgical procedures and outcomes. The use of this technique allows the statistically sound identification of distinct subgroups using a combination of categorical, ordinal, and continuous descriptive variables. Best suited to large data sets we feel this technique was useful in analyzing the data presented in this paper and has additional applications within large surgical datasets.
Use of adjuvant triple medical therapy consisting of a proton pump inhibitor, anti-inflammatory systemic antibiotic, and inhaled corticosteroid in Cluster 2 may account for the observed treatment benefit. While the pathophysiology of iSGS remains incompletely understood, several theories have been presented ^9–11^. Recent work supports the concept that an impaired subglottic epithelial barrier allows displacement of a normal microbiome into the lamina propria, driving sustained host adaptive immune activation and subsequent tissue remodeling ^12^. Mechanistically, the three drug post-operative regime may address several of these components. Proton pump inhibitors (PPIs) may limit the mucosal damage from refluxed gastric acid during the initial healing phase after dilation.
Recent literature also demonstrates a large inflammatory cellular infiltrate in the affected mucosa ^13^. This inflammation may in part be displacement of the native microbiome across a dysfunctional epithelial barrier into the lamina propria ^12^. TMP-SMX has been previously demonstrated to limit disease relapse in granulomatosis with polyangiitis (GPA), specifically reduction in airway disease flares ^14^. This effect may be a reduction in triggering organisms or via an immunomodulating effect ^14^. While there has been conflicting research on the effect of TMP-SMX on iSGS subglottic stenosis outcomes ^15^, alternate data support an effect of TMP-SMX on reducing recurrence risk following endoscopic laser wedge excision ^16^. Inhaled corticosteroids may also blunt the observed inflammatory response. Yet, similar to PPI, data on the impact of isolated adjuvant inhaled corticosteroids on time to recurrence or peak expiratory flow have not demonstrated a significant effect ^15^.
Given the nature of our study, our sample size may have been too small to detect significant differences across clusters. We excluded centers reporting fewer than 10 EDs for iSGS. This ensured sufficient patient data to perform subgroup analyses but limited the overall number of centers included in our study thus limiting the power of this study. Another limitation is the time difference between when the patient data and the procedural data were collected; during this 4–6 year gap, it is possible that there may have been slight variations in procedural techniques. Additionally, the procedure-specific variables were defined by consensus-based expert opinion; additional explanatory variables, not considered when initially designing the survey instruments, may exist. Yet there is evidence the variables captured are robust; qualitative responses from each institution did not identify additional themes not explicitly covered in the initial survey. Because institution-level data was self-reported, slight deviations from case to case may not be reflected in our data. Additionally, variability in patient populations between centers may account for some differences in time to recurrence regardless of procedural variables. Cluster 2 also had a higher percentage of peri/post-menopausal patients (57% postmenopausal vs 30% and 20%; p = 0.033). It is possible that lower levels of circulating estrogen may account for the observed differences in clinical outcome after ED, rather than procedural variables.
Both outcome and procedural technique vary among centers employing ED to treat iSGS. A transient delay in recurrence was observed among centers that routinely prescribed adjuvant medical therapy (antibiotic, inhaled corticosteroid, and PPI) to iSGS patients after endoscopic dilation. Both patient-reported data and peak expiratory flow data also appear to support the benefit of triple therapy in delaying disease recurrence. Prospective studies are needed to understand the effects of adjuvant medical therapy on recurrence after endoscopic dilation.