The Study of Laryngoscopic and Autonomic Patterns in Exercise-Induced Laryngeal Obstruction

Abstract

Objectives:

(1) Identify laryngeal patterns axiomatic to exercise-induced laryngeal obstruction (EILO) and (2) investigate the role of autonomic function in EILO.

Methods:

Twenty-seven athletic adolescents (13 EILO, 14 control) underwent laryngoscopy at rest and exercise. Glottal configurations, supraglottic dynamics, systolic blood pressure responses, and heart rate recovery were compared between conditions and groups.

Results:

Inspiratory glottal angles were smaller in the EILO group than the control group with exercise. However, group differences were not statistically significant (P > .05), likely due to high variability of laryngeal responses in the EILO group. Expiratory glottal patterns showed statistically greater abductory responses to exercise in the control group (P = .001) but not the EILO group (P > .05). Arytenoid prolapse occurred variably in both groups. Systolic blood pressure responses to exercise were higher in the control group, and heart rate recovery was faster in the EILO group. However, no significant differences were seen between the 2 groups on either autonomic parameter (P > .05).

Conclusions:

“Paradoxical” inspiratory and blunted expiratory vocal fold pattern responses to exercise best characterize EILO. Group differences were only seen with exercise challenge, thus highlighting the utility of provocation and control groups to identify EILO.

Introduction

Inducible laryngeal obstruction (ILO)—also known as paradoxical vocal fold motion (PVFM) disorder and vocal cord dysfunction (VCD)—is a consensus term^1,2 used to describe a spectrum of upper airway disorders thought to originate from paroxysmal constriction of the laryngeal muscles. Episodes are typically triggered acutely by environmental or systemic stimuli.^3–6 Laryngeal kinematics in ILO have variably been described as prolapse, narrowing, constriction, and obstruction⁶; the arytenoids, aryepiglottic folds, epiglottis, ventricular folds, and true vocal folds have all been implicated. However, whether these patterns are indicative of ILO is currently unknown due to methodological constraints involving lack of comparative control groups and dearth of validated methods to determine diagnostic benchmarks for pathology.^7,8

These gaps also challenge the clinical utility of laryngoscopy as the current “gold standard” to diagnose ILO. Without validated objective methods and benchmarks to help differentiate laryngeal pathophysiology from normal variations in laryngeal responses to respiratory perturbations, practitioners are left to rely on clinical intuition to diagnose ILO. Unfortunately, current practices have led to both under- and overdiagnosis of the condition.¹ It is therefore not surprising erroneous diagnosis initially occurs in up to 90% of individuals,^9,10 which is especially troubling for the 5.7% to 7.5% of adolescents with one ILO variant—exercise-induced ILO (EILO).^11,12 Inaccurate diagnosis in the adolescent EILO population can lead to iatrogenic consequences in otherwise healthy young individuals, places heavy financial and resource burdens on the family and medical system, and can result in loss of scholarships and other athletic opportunities.¹⁰ Although a recent transatlantic panel provided comprehensive diagnostic recommendations for the diagnosis of EILO, highlighting the clinical utility of continuous laryngoscopy with exercise (CLE) (for details, see Røksund et al¹), they also acknowledged the subjectivity of current approaches to diagnose ILO.¹ Therefore, to begin to address these gaps, the first aim of the study was to determine laryngeal patterns indicative of ILO, starting with the EILO variant. Typical versus atypical laryngeal responses were identified, quantified, and directly compared across 2 groups of athletes with and without EILO.

In addition to these gaps, there is also a poor understanding of underlying mechanisms driving clinical expression in ILO.⁷ Previous literature has alluded to altered autonomic nervous system (ANS) balance as the cause;¹³ recent literature has also shown increases in intrinsic laryngeal muscle activity with simultaneous autonomic arousal in healthy subjects.^14,15 Since the sympathetic (SNS) and parasympathetic nervous systems (PNS)—the 2 subsystems of the ANS—work together to regulate cardiovascular parameters, heart rate and blood pressure have previously been used as noninvasive proxy methods to measure ANS activity.^16–18 More specifically, the SNS is primarily responsible for mediating changes in systolic blood pressure (SBP) from rest to rigorous exercise,^16,19 while heart rate recovery (HRR)—the difference between heart rate at peak exercise and 2-minutes post-exertion—is thought to be a robust indicator of sympathetic withdrawal and simultaneous parasympathetic reactivation needed for homeostatic balance.^19–22 Although aberrant sympathovagal involvement has been established in conditions that co-occur with ILO (eg, asthma, anxiety),^13,20,23 the role of ANS in laryngeal breathing pathology has yet to be determined. Therefore, the second objective of the present study was to determine the plausibility of autonomic dysfunction as a mechanism in EILO etiology via pilot data.

Materials and Methods

Participants

Twenty-seven adolescent athletes (13 EILO, 14 controls) were recruited for the study. Young competitive athletes were chosen since this cohort most often presents with the EILO variant in clinical settings. Patients with suspected EILO—characterized as dyspnea of unknown etiology causing decrements to athletic performance—were approached for written consent at their initial consultation at the Voice and Speech Laboratory at Massachusetts Eye and Ear (MEE). All participants in the EILO group had undergone previous extensive workup to rule out pulmonary, gastrointestinal, and cardiovascular conditions as causes of their dyspneic symptoms. Control participants from local sports teams and schools in the New England region were consented and enrolled at the time they presented to the clinic for possible participation. General inclusion criteria across groups included an age range of 12 to 18 years and participation in competitive (intramural or extramural) aerobic athletics at least 3 times a week for a minimum of 40 minutes per session. To be eligible for the experimental (EILO) group, participants needed to score greater than a 10/40 on the Dyspnea Index (DI), a symptom-severity questionnaire previously validated in patients with upper airway pathologies.²⁴ Athletes were eligible for the control group if they experienced no dyspneic symptoms detrimental to athletic performance within the preceding 6 months, scored less than a 7/40 on the DI,²⁵ and had no medical history of neurological, cardiovascular, or pulmonary disorders. Exclusionary criteria for both groups included organic lesions obstructing more than a third of the upper airway and intolerance to flexible laryngoscopy.

Procedures

Laryngoscopy protocol.

Once consented, participants were fitted with a heart rate monitor (Polar H7 Bluetooth Heart Rate Sensor & Fitness Tracker, Kempele, Finland) to measure continuous heart rate, and blood pressures were measured 3 times consecutively with a pressure cuff (Welch Allyn Spot Vital Signs LXi, Skaneateles Falls, New York, USA). Participants were trained to use a 1 to 8 physical effort scale for indication of current physical effort (1 = no exertion, 8 = maximum exertion), using comparable concrete examples for each exertion level (eg, reading a book = 1, finishing kick of a race = 8). Participants were then asked to rate their present dyspnea and leg fatigue levels on a 1 to 100 continuous visual analog scale (VAS). Participants were informed they would be encouraged to pedal on a stationary trainer (SF-B1203 Indoor Cycle Trainer, Sunny Health and Fitness, Los Angeles, California, USA) as hard as they could until they reached an 8 on the scale.

Oxymetazoline was sprayed into the nostrils to decongest the nasal passages. Lubricating surgical jelly was placed near the tip of the flexible nasoendoscope (KayPENTAX EPK-1000, Kay Elemetrics Corp, Lincoln Park, New Jersey, USA) and passed through the nare with the larger meatus. Once inserted, baseline laryngeal movements were video recorded using halogen light (nStream G3, Image Stream Medical, Littleton, Massachusetts, USA) for later analysis. Ribcage movement was monitored visually, and nasal airflow emissions were monitored via tactile feedback on the same hand used to drive the tip of the endoscope to identify respiratory cycle boundaries. Any respiratory-laryngeal pattern asynchrony was verbally documented on the video and later used during data extraction. Normal structural and organic laryngeal anatomy landmarks were also confirmed. During acquisition of laryngeal patterns at rest (baseline), 3 consecutive blood pressure measurements were taken for the second time.

Participants were then asked to pedal continuously on the trainer between 50 and 110 rpms while resistance was incrementally increased every 30 seconds. Simultaneous laryngeal visualization was conducted in real time to observe laryngeal responses to exertion. Laryngeal-respiratory synchronization was monitored simultaneously with the exercise challenge using the same methods as the baseline laryngoscopy condition. Participants in the control group pedaled on the trainer with continuous laryngoscopy until they reported an 8 on the exertion scale (maximum exertion), at which point 30 additional seconds were recorded before the exercise challenge was stopped. Participants in the EILO group cycled with continuous laryngoscopy until either an 8 on the exertion scale was reported or an EILO episode was provoked, at which point an additional 30 seconds was recorded before the protocol was terminated. Positive EILO episodes were based on self-reported symptoms (eg, dyspnea) and simultaneous anterior-posterior/lateral-medial glottic or supraglottic pattern changes from rest. If an EILO episode could not be elicited, the exercise challenge was terminated after 1 minute at maximum exertion. Immediately at termination of the exercise challenge, cardiovascular parameters were taken for the third and final time. Participants then rated the severity of their dyspnea and leg fatigue experienced at the apex of the exercise challenge on the same 1 to 100 scale.

Data Extraction and Analysis.

For each participant, 30-second videos at baseline laryngoscopy and the apex of maximum exertion (or EILO episode) with laryngoscopy were extracted into 2 separate MP4 files. Eight digital still images per participant were also acquired: 2 at the peak of inspiration and 2 at the peak of expiration for each of the 2 laryngoscopy conditions. Five independent raters blinded to group and condition identified anterior glottal angles on the digital still images using a customized MATLAB program (Figure 1; see McKenna et al²⁶ for details) and rated levels of supraglottic activity on the videos using previously validated 0 to 3 severity scales (see Table 1 for details). Raters were either fellowship-trained laryngologists or certified speech-language pathologists with at least 5 years of laryngoscopy experience. One of the raters conducted analysis on all the images and videos. The other 4 raters completed analysis on 30% of the images and 30% of the videos, randomized to each rater, to confirm previously validated interrater reliability (see Shembel et al⁶ for details). Prior to evaluating respective images and videos, raters each trained on randomized 20 images and 8 videos.

Figure 1.

Anterior glottal angle measurements. Raters first identified the medial edge of each vocal fold. Raters then placed a coordinate at the medial tip of the right vocal process with a single click of the mouse, followed by double clicking the mouse at the vertex of the anterior commissure on the ipsilateral side (or slightly past the commissure) to create a line that followed the medial edge of the superior vocal fold. The same patterns were conducted on the left vocal fold, creating crosshairs at the anterior commissure. Anterior glottal angles were calculated from the crosshairs and automatically populated into an Excel spreadsheet using the customized MATLAB program.

Table 1.

Rating System for Supraglottic Laryngeal Responses.^a

Obstruction Type	0	1	2	3
Arytenoid prolapseb	Expected maximal abduction of the aryepiglottic folds with no visible medial rotation (top of cuneiform tubercles pointed vertical or slightly lateral)	Visual medial rotation of the cranial edge of the aryepiglottic folds and tops of the cuneiform tubercles	Further medial rotation of the cuneiform tubercles with exposure of the mucosa on the lateral sides of the tubercles	Medial rotation until near horizontal position of the cuneiform tubercles and tops of the cuneiform tubercles move toward the midline
Epiglottic collapsec	Epiglottis hugs the tongue	Epiglottis between 0° and 45°	Epiglottis between 45° and 90°	Epiglottis over 90°
Ventricular (false) fold compressiond	Absent	One-third of the true vocal folds covered	Two-thirds of the true vocal folds covered	True vocal folds completely hidden

^aThe rating scale was used to quantify severity of supraglottic obstruction across 3 supraglottic patterns: arytenoid prolapse, epiglottic collapse, and ventricular compression. The severity of each loci of obstruction was rated on a 0 to 3 scale (0 = none, 1 = mild, 2 = moderate, 3 = severe). Descriptions representing severity levels of each parameter was provided to each rater (see previous work by Maat²⁷, Catalfumo et al²⁸, and Schönweile et al²⁹ for details on validation of supraglottic rating methods).

^bValidated by Maat (2011).

^cValidated by Catalfumo et al. (1998).

^dValidated by Schonweile et al (1998).

Interrater Reliability and Post hoc Analysis.

Once raters analyzed the images, intraclass correlation coefficients (2-way mixed, absolute agreement) for each anterior glottal angle across 100% of images analyzed by rater 1 and 30% for each of the other four raters were run to confirm precision of angle estimates (average = 0.97, SD = 0.02; range, 0.9180-.995). Glottal angles were then averaged across the raters, and the magnitude of differences between each set of images within the same laryngoscopy condition (baseline, exercise) and respiratory cycle (inhale, exhale) was examined with boxplots. Frame-by-frame analysis was conducted on boxplot outliers to further evaluate respiratory-laryngeal pattern differences.

Autonomic calculations.

Changes in systolic blood pressure from rest to maximum exertion and heart rate recovery, calculated as the highest heart rate at maximum exertion subtracted from heart rate 2 minutes post-maximum exertion, were determined for each participant and used as outcome variables.

Statistical analysis.

Normality and variance assumption for the laryngeal parameters were violated. Therefore, Kruskal Wallis H tests were conducted to determine group differences (EILO, control), and Wilcoxon sign-rank tests were used to determine condition differences (baseline, max-exercise) for glottal angles and supraglottic responses. Data for both systolic blood pressure change (ΔSBP) and 2-minute heart rate recovery (2-HRR) were normally distributed (as assessed by Shapiro-Wilk’s test, P > .05), and assumption of homogeneity was met (as assessed by Levene’s test for equality of variances, P > .05). Therefore, independent samples t tests were performed between groups as a function of condition, with ΔSBP and 2-HRR as the dependent variables, respectively. All parametric and nonparametric statistical analyses were corrected with Bonferroni adjustments to account for Type 1 errors.

Results

Participant Characteristics and Demographics

Results showed groups were comparable in age, gender, and athletic level.¹ Specifically, 9 females and 4 males (age: 14.54 years ± 2.03 years) were recruited into the EILO group and 9 females and 5 males (age: 16.87 years ± 1.19 years) into the control (CONT) group. All participants reached an 8 on the exertion scale, and both groups reported similar levels of leg fatigue with baseline (EILO: 5.54, CONT: 4.79, P > .05) and exercise laryngoscopy (EILO: 57.54, CONT: 63.43, P > .05). Dyspnea in patients with EILO were twice as high (M = 62.90/100) as controls (M = 33.50/100) during strenuous exercise (P = .01). Time to maximum exertion for the exercise challenge (8 on 1/8 scale) was 277.00 ± 85.87 seconds in the EILO group and 265.71 ± 48.48 seconds in the control group. All 27 participants tolerated laryngoscopy. Episodes of EILO were induced with exercise challenge in all but 1 participant in the EILO group, and data for that individual (male, 17 years) were excluded from statistical analysis.

Glottal Configuration

Glottal Angles (Inspiration).

Inspiratory glottal angles were similar between groups at baseline (Figure 2). In contrast, average inspiratory angles decreased from baseline to exercise in the EILO group. Although descriptive findings showed adduction occurred with inspiration in response to exercise in the EILO group, these differences were not statistically significant across condition (P = .05) or group (P = .15), likely due to high variability in laryngeal responses in the EILO group. These variable responses were evidenced by the twofold standard deviation increase from baseline to exercise (Figure 2) as well as large magnitude angle differences (Figure 3A) and variable laryngeal responses across the same condition in some but not all EILO participants (Figure 3B).

Figure 2.

Anterior glottal angle (inspire and expire) configurations. The figure represents both an average of 2 images taken within the same condition and group averages. Glottal angles at baseline were similar between the 2 groups. In contrast, robust group differences were seen with exercise challenge. Participants in the exercise-induced laryngeal obstruction (EILO) group showed decreased inspiratory glottal configurations with exercise (ie, paradoxical vocal fold motion). Although expiratory glottal angles from rest to exercise challenge increased in both groups, angles were larger in the control group. Reduced abductory patterns during the expiratory phase in the EILO group, compared to the control group, may indicate a sluggish expiratory response to exercise.

Figure 3.

(A) Anterior glottal angle differences between 2 images taken within the same condition (baseline exhale, baseline inhale, exercise exhale, exercise inhale). This figure represents the level of variability in angle responses that occurred within the same condition. Specifically, the majority of angle differences were within 15° of each other. However, greater inspiratory differences were seen in 2 participants in the exercise-induced laryngeal obstruction (EILO) during strenuous exercise (outliers: E01 and E07). (B) Post hoc analysis of all angles within the same condition for the 2 individuals showed high variability in inspiratory laryngeal responses across the same condition (represented by line graphs). However, this variability was not present in all individuals in the EILO group; see line graph for participant E10 for example. Angles measured across the same condition were also less variable in the control group; see C06 and C10 for comparison.

Glottal Angles (Expiration).

Average expiratory glottal angles were also similar between the EILO and control groups at rest (Figure 2). Although increases in expiratory glottal angles were seen in both the EILO and control groups from baseline to exercise, the size of the angle (eg, amount of abduction) was greater in the control group. Differences in expiratory angles from baseline to exercise were significant in the control group (P = .001) but not the EILO group (P > .05).

Supraglottic Laryngeal Patterns

Interestingly, exercise had a significant effect on arytenoid response in both the EILO (P = .002) and control (P = .01) groups. However, there were no differences between groups at rest (P = .83) or exercise (P = .22). Of note, when arytenoid prolapse was present in control participants, individuals were completely asymptomatic (eg, no dyspnea). Other supraglottic patterns showed epiglottic and ventricular patterns were unremarkable for both group and condition (P > .05) (see Table 2 for prevalence of supraglottic patterns at baseline and exercise conditions).

Table 2.

Descriptive Findings for Supraglottic Patterns at Baseline (Rest) and Maximum Exertion (Exercise Challenge) in Participants With EILO and Controls.^a

		EILO (%)				Control (%)
Supraglottic Patterns		0 (None)	1 (Mild)	2 (Moderate)	3 (Severe)	0 (None)	1 (Mild)	2 (Moderate)	3 (Severe)
Arytenoid prolapse	Baseline	75	25	0	0	79	21	0	0
	Exercise	8	42	50	0	21	50	29	0
Epiglottis collapse	Baseline	67	33	0	0	78	14	7	0
	Exercise	83	17	0	0	79	21	0	0
Ventricular compression	Baseline	100	0	0	0	100	0	0	0
	Exercise	83	17	0	0	79	21	0	0

^aPrevalence is based on the 0 to 3 severity rating scale (cf. Table 1). EILO, exercise-induced laryngeal obstruction.

Cardiovascular Biomarkers

Average ΔSBP was higher in the control group than the EILO group (Figure 4A). These patterns contradict previous literature suggesting higher sympathetic “fight or flight” stress responses common to individuals with EILO.^30–33 Although control participants, on average, had higher sympathetic responses to exercise, these differences were not statistically significant, t(22) = −0.50, P = .63, d = 0.20. Conversely, 2-HRR was greater in the EILO group (M = 54.08, SD = 15.10) than the control group (M = 49.15, SD = 15.91) (Figure 4B). However, differences were not statistically significant, t(24) = 0.81, p = .43, d = 0.32.

Figure 4.

Parasympathetic and sympathetic autonomic responses between the 2 athletic groups, determined by (A) magnitude of change in systolic blood pressure from rest to maximum exertion and (B) 2-minute heart rate recovery (2-HRR). Changes in pressures were slightly higher in the control group than exercise-induced laryngeal obstruction (EILO) group while 2-HRR was slightly faster in the EILO group, compared to the control group. Although group differences were not significant, trends in these patterns might suggest sluggish sympathetic responses to exercise or heightened parasympathetic activity, at least in some individuals with EILO. Therefore, the role of autonomic responses in EILO warrants future investigations.

Discussion

Results of this study both parallel and contradict previous reports of purported EILO presentations. The 15% average decrease in inspiratory glottal angles from rest to strenuous exercise seen in the study’s EILO group align with previous subjective descriptions of “paradoxical” inspiratory vocal fold adduction, with more than an >8° decrease seen in 9 out of 12 EILO participants (baseline: M = 53.77° ± 10.32°; exercise: M = 40.15° ± 20.85°). There was high variability in inspiratory laryngeal response patterns across participants in the EILO group, which likely contributed to null differences between groups (P = .15) and across conditions (P = .05). In addition to high intersubject variability, there was also high intrasubject variability within the same condition and respiratory cycle patterns in some individuals (cf. Figure 3). Heterogeneity in clinical patterns has previously been described by Røksund and colleagues,¹ who attributed these varied patterns to the complexity of laryngeal neuroanatomy and physiology; varied laryngeal patterns also suggest more than one mechanism underlying EILO clinical expression. In contrast to these “classic” inspiratory patterns thought to indicate EILO, reports of expiratory laryngeal patterns have traditionally remained more elusive.^34–39 We found increases in expiratory glottal angle size in response to exercise to be smaller in the EILO group (11% average increase from rest) compared to the control group (36% average increase from rest). Smaller glottal size from rest to exercise in EILO are suggestive of blunted abductory responses and expiratory inefficiency; this is in contrast to the counterproductive inspiratory adduction “paradox” seen in EILO.

With regard to supraglottic findings, arytenoid prolapse has previously been implicated as a diagnostic indicator of EILO.^3,40–42 However, due to lack of empirical study, it was unknown whether these patterns could also occur in young athletes without EILO. Our results suggest severity of arytenoid prolapse occurs variably in young athletes during strenuous physical activity. Prolapse of the arytenoids could have more to do with increased laryngeal cartilage pliability and smaller airways common to juveniles. When combined with high negative inspiratory pressures during strenuous exercise, supraglottic structures can prolapse into the laryngeal lumen.⁴³ Therefore, arytenoid prolapse as a diagnostic indicator of EILO in and of itself should be used with caution. Finally, the low prevalence of epiglottic and ventricular responses to exercise in the EILO group suggest these latter 2 patterns may reflect other ILO variants (e.g., irritable larynx syndrome^44,45) if they do, in fact, play a role in ILO pathoetiology.

These study findings highlight the importance of control groups and utility of provocation in study designs. Specifically, results showed remarkably similar glottal configurations between groups at rest for both inspiration (~53°) and expiration (~36°). Group differences became evident only with strenuous exercise (eg, average inspiratory angles: 40° in EILO group vs 54° in control group). Similarities in arytenoid patterns between groups further support the need for controlled comparisons so as not to “pathologize” laryngeal responses to respiratory perturbations. Stated differently, the use of a provocation challenge in the context of comparison groups can differentiate the “normal” from the “abnormal” and set the foundation for improvements in diagnostic approaches.

Sympathovagal responses also showed interesting trends between groups. Despite seemingly alarming “fight or flight” presentations characteristic of EILO, in which a greater sympathetic response would be anticipated, the opposite occurred. Participants with EILO showed, on average, smaller ΔSBP and faster 2-minute post-exertion HRR. These findings could point to a more sluggish sympathetic exercise response or heightened parasympathetic activity in EILO. This is not to say the ANS does play a role in EILO. Further investigations with larger cohorts and more involved methodology are needed to directly determine the role of ANS in the context of EILO. However, preliminary study trends are interesting and should be further explored.

Conclusion

Study findings have several key implications. First, results illustrate the importance of normative comparisons and provocation challenges to better characterize and standardize laryngeal response patterns and prevent over-attribution of laryngeal responses to pathology. Second, results demonstrate paradoxical adductory inspiratory patterns and blunted expiratory abduction to be characteristic of EILO. However, heterogeneity among these patterns can also be expected both within and across patients with EILO—despite this seemingly homogenous athletic group—and may reflect disparate pathoetiological mechanisms. Finally, trends in autonomic responses may point to altered autonomic function, at least in some individuals with EILO; however, further study is needed.