An ex-vivo model examining acoustics and aerodynamic effects following medialization with and without arytenoid adduction

Abstract

Objectives/Hypothesis:

Quantify differences in acoustics and intraglottal flow fields between Thyroplasty Type 1 (TT1) with and without arytenoid adduction (AA) using excised canine larynx model.

Study Design:

Basic science experiments using excised larynges.

Methods:

Surgical procedures were implemented in 8 excised canine larynges. Acoustics and intraglottal flow measurements were taken at low and high subglottal pressures in each experimental setup.

Results:

In all larynges, vocal efficiency (VE) and cepstrum peak prominence (CPP) were higher, and the mean phonatory flow rate was lower in TT1 with AA than without AA. The glottal asymmetry is reduced with AA and promotes the formation of stronger vortices in the glottal flow during the closing phase of the vibrating folds.

Conclusions:

Findings suggest a clear acoustic and aerodynamic benefit to the addition of AA when performing TT1. It shows significant improvement in CPP, translating to decreased breathiness and dysphonia and increased VE, leading to easier and more sustainable phonation. Stronger intraglottal vortices are known to be correlated with the loudness of voice produced by phonation.

Level of Evidence:

NA

INTRODUCTION

Unilateral vocal fold paralysis (UVFP) occurs due to decreased residual innervation to one true vocal fold, resulting in paralyzed vocal fold movement. Patients with unilateral vocal fold paralysis complain of a soft, breathy voice that is difficult to understand in noisy environments. Often UVFP is accompanied by increased vocal fatigue and structural asymmetries in the glottis. Malposition of the arytenoid cartilage occurs because of differences in forces that affect the homolateral laryngeal muscles after damage and pathological reinnervation, resulting in significant asymmetries in vocal fold length, height, and tension.^{1 2}

UVFP treatments aim to permanently move the paralyzed vocal fold medially, so the adductory motion of the non-paralyzed fold will result in the prephonatory glottal shape (i.e., closed glottis). The approximation of the paralyzed fold is made using either injection or thyroplasty. These phonosurgical approaches improve laryngeal function even if complete glottal closure is not achieved because they reduce the glottal gap.

A common treatment for UVFP is Thyroplasty Type 1 (TT1)³. TT1 procedure uses an implant to medialize the membranous paralyzed fold to improve glottal closure.^4,5 Surgically, there remains controversy on how to restore optimum vocal function when performing TT1. No optimal technique has been found, and the revision rate has been reported as high as 12–25%.^1,6 Debate regarding TT1 procedures includes optimum implant size, shape, location, and whether to include arytenoid adduction (AA).^7–11

AA uses a surgical stitch that, when used in conjunction with TT1, can further medialize the fold, close the posterior gap (sometimes referred to as chink), and minimize height asymmetries and medial-lateral asymmetries between the folds.¹² The addition of AA can be technically difficult, and there is disagreement about the necessity of this stitch. Most often, surgeons include this stitch when a large posterior gap remains following TT1 to further medialize the paralyzed fold, although disagreement remains among surgeons as to how large a posterior gap warrants AA.¹³

Although many studies discuss either TT1 or AA, few compare TT1 only to TT1 in conjunction with AA. Some studies found no difference in voice outcome between patients treated with TT1 and TT1 with AA^14,15, while others found an improvement in voice outcome when AA was included.^16–18 Studies that compared aerodynamic measures found that TT1 in conjunction with AA showed a greater reduction in mean phonatory flow rate (Q), which translates to increased vocal efficiency, and a larger increase in mean phonatory time than TT1 only.^6–8 However, the main limitation of these studies is that they were based on retrospective clinical data instead of a prospective study design. As such, understanding the benefits of adding AA to TT1 is limited because there was no control on how patients were assigned to each group. Specifically, there was no consistency in the patients’ etiology that guided the surgical decision on whether or not to include AA with TT1.

Modeling study is the ideal tool for understanding how including AA with a TT1 procedure would affect the surgical outcome in the same larynx. Such study design is not feasible in clinical settings. Models based on excised canine larynges^18,19 have shown a decrease in shimmer and jitter with the addition of AA. However, the statistical significance for these acoustic features was not shown¹⁸, or it was not significant.¹⁹ The latter study, however, did show a statistically significant difference in aerodynamic measures. Specifically, the addition of AA resulted in a reduction in phonation threshold flow rate, and phonation threshold power (phonation threshold flow multiplied by the phonation threshold pressure). They also observed statistically significant difference in the size of the posterior gap and vocal folds asymmetry between TT1 only and TT1 with AA.

Vocal fold asymmetries in their left-right position, height, and length were shown to affect the intraglottal flow characteristics, as well as acoustics features, in the excise canine larynx model.²⁰ This study found that the strength of the intraglottal flow separation vortices (FSV) was significantly reduced in the presence of vocal asymmetries. The FSV form near the superior aspect of the folds when the intraglottal flow separates from the divergent wall of the glottis. It was shown both experimentally^21,22 and computationally²³ that the strength of these vortices is correlated with the maximum flow declination rate (MFDR) of the glottal flow. Clinically, MFDR is important because it is highly correlated with the sound pressure level (SPL), the acoustic energy in the higher harmonics, and the vocal efficiency.^24–26 Because AA reduces glottal asymmetry, it is reasonable to expect that it will also affect the formation of intraglottal vortices.

The current study aims to quantify differences in the intraglottal flow field and the acoustics between TT1 and TT1 with AA using excised canine larynx model. Our hypothesis is that TT1 with AA will increase the strength of the FSV and improve objective measures of voice outcomes compared with TT1 alone in the excised canine larynx model.

METHODS

Experimental setup.

Eight excised canine larynges (L1-L8) were harvested immediately after the animals were euthanized. All cartilage and soft tissue above the vocal folds were removed to gain an unobstructed view of the vocal folds. The tracheas were kept 3–5cm long and placed on an aerodynamic nozzle that supplied humidified, conditioned airflow to the glottis. The larynx was positioned using three-prong support inserted into the arytenoid’s lateral surface and used for medial-lateral translation of the vocal processes.

Phonation was induced by controlling the conditioned airflow supplied to the larynx. The mass flow rate (Q) was controlled and measured using a flow meter (MicroMotion Inc, CMF025 Coriolis Flow Meter). The static pressure inside the nozzle was measured using a pressure transducer (Honeywell, FPG) and was used as the value for the subglottal pressure, Psg. Acoustic measurements were taken using a 0.5-inch free-field microphone (model 4950, Bruel &Kjaer). The microphone was placed approximately 30 cm laterally and superiorly to the glottal exit. The microphone, pressure, and flow rate data signals were sampled at 40 kHz and were synchronized using a data acquisition system (PXIe-6356, National Instruments). The schematic for the experimental apparatus is shown in Figure 1a.

Figure 1-.

a) Schematic of the experimental apparatus. b) Difference in prephonatory glottal shape between baseline and the surgical cases. The paralyzed vocal fold is simulated on the right side.

Measurements in each larynx were taken at low and high Psg in each experimental configuration. Low Psg was defined as 2–5 cmH₂O above the phonation threshold pressure (PTP). High Psg was defined as 5–10 cmH₂O above the low Psg value for the same condition. The experimental configurations for each larynx included baseline, TT1 alone, and TT1+AA. Baseline cases were taken before the surgical manipulation by adducting both folds in the midline, simulating normal vibrations. The surgical manipulations were done ipsilateral to the “paralyzed” fold while the “non-paralyzed” fold was adducted medially by translating its vocal process (i.e., same as in baseline). The measurements were first taken with TT1 alone in all larynges before adding the AA.

Surgical Techniques.

Medialization of the paralyzed vocal fold was achieved by placing an implant through a thyroplasty window cartilage in the thyroid cartilage. The implant was carved out of silicone material deemed to have a similar stiffness to the Silastic implant commonly used in the TT1 procedure. The optimal degree of medialization was determined as adducting the fold until the vocal processes are just touching, which was confirmed visually by the co-authors (S.K and R.H).

AA was performed on the paralyzed fold according to the procedure described by Isshiki et al.²⁷ A suture was passed through the muscular process of the arytenoid and exited above and below the anterior cricoid; these ends were tied together and the suture was then tightened to rotate the arytenoid and adduct the fold. Care was taken not to over-tighten the suture and avoid hyperadduction of the arytenoid. The difference in prephonatory glottal shape between baseline and the simulated phonosurgery is shown in Figure 1b.

Intraglottal flow measurements.

In three excised larynges (L6-L8), intraglottal flow velocity measurements were taken using phase-locked particle image velocimetry (PIV). The description of the methodology for measuring the intraglottal flow using PIV can be found in Oren et al.²⁸, and the details about phase-locked measurements can be found in Jiang et al.²³ In short, intraglottal velocity measurements were taken by illuminating the coronal plane with a laser sheet, halfway between the vocal process and the anterior commissure. The PIV camera was positioned above the vocal folds at an oblique angle that enables visualizing the glottal jet. Phase information was determined using a waveform signal of the glottal cycle from an electroglottograph (EGG). The phase of the glottal cycle was defined by θ varying from θ = 0° or 360° at the beginning of superior edge opening. Phase-locked PIV data was collected every 5 degrees during the closing period of the vibrating folds, with 30 images collected at each phase.

Aerodynamic and acoustic analyses.

The strength of each intraglottal vortex was quantified using its circulation value. Circulation is a measure for the amount of rotation within an enclosed region of the flow field: $\Gamma ={\iint }_s\omega ds$, where ω is the vorticity normal to the measurement plane, and s is the area encompassing the vortex. Description of how the closed region around FSV was determined can be found in Farbos de Luzan, et al.²⁹

Analyses of acoustic features include vocal efficiency (VE) and cepstrum peak prominence (CPP). VE was calculated as the ratio of the acoustic power to the aerodynamic power, detailed by Schutte.³⁰ In this classic definition, VE=2πr²10⁻¹²10^SPL/10/PsgQ, where r is the distance from the microphone to the sound source (glottal exit) and SPL is the sound pressure level. Clinically, the VE translates to easier and more sustainable phonation. CPP was used in the current study because it showed a strong (negative) correlation with breathiness and dysphonia^31–33, two perceptual measures that characterized patients with UVFP.

Statistical Analysis.

Each experimental configuration was analyzed for statistical significance. A fixed-effect model was used to assess the association of the dependent variables, VE, Q, and CPP, to the fixed effects of Psg and experimental configurations. The statistical analysis was done using SAS 9.3 software (SAS, Gary, NC) with a two-sided test set at a 5% significance level (p = 0.05). Using 8 excised larynges gave 80% of power to detect an effect size of 1.

RESULTS

Adding AA to TT1 affected the acoustics and aerodynamics parameters used to assess the phonation process. The measured data and their statistical analysis are summarized in Tables 1 and 2, respectively. In all 8 larynges, VE and CPP were higher, and Q was lower in TT1 with AA than without AA. The differences in Q, CPP, and VE were also statistically significant, but the changes in VE were more pronounced at high Psg than low Psg (P=.00002 and .004, respectively). These results are similar to the findings of Green et al.¹⁸

Table 1 –

Flow and acoustic data.

Larynx	TT1					TT1+AA
	Psg (cmH 2 O)	Q (lpm)	Vocal efficiency	CPP (dB)	Γmax (1/s)	Psg (cmH 2 O)	Q (lpm)	Vocal efficiency	CPP (dB)	Γmax (1/s)
L1	17.0	50.9	.054	15		14.5	28.8	.058	16.3
	21.9	71.5	.075	15.9		23.2	59.8	.088	16.1
L2	15	63	.018	8.4		15.2	45.2	.046	14.3
	24.3	126.3	.062	11.7		22.7	79.2	.086	16.9
L3	16.9	63.3	.038	10.6		15.5	23.5	.052	13.0
	19.8	63.4	.038	10.5		20.7	30.3	.106	13.0
L4	9.8	62.1	.052	12.8		11.5	31.4	.103	15
	13.6	105.2	.048	11.5		13.2	58.24	.097	13.2
L5	11.3	77.8	.018	10.1		14.3	14.9	.073	15.7
	14.1	105.6	.022	9.7		18.3	20	.099	13.3
L6	14.2	103.4	.053	12.6	9.03e4	14.4	67.9	.071	14.9	2.34e5
	18.8	67.9	.078	13.6	9.34e4	18.7	78.3	.092	13.9	2.69e5
L7	13.7	73.5	.035	9.2	2.80e4	14.7	35	.074	13	3.70e4
	18.7	98.4	.045	9.7	4.46e4	19.9	43.5	.096	14.6	5.25e4
L8	16.8	60	.046	9.0	1.53e4	16.3	18.6	.104	10.6	1.56e5
	21.0	79.6	.053	8.2	2.59e4	21.8	30.6	.093	11.9	1.94e5

Table 2 –

Summary of statistical analysis. Mean values ± standard deviations are shown.

Parameter		N	TT1	TT1+AA	P-value
VE	low Psg	8	.039±.01	.073±.02	.004
	high Psg	8	.053±.02	.095±.01	.00002
	combined	16	.046±.02	.082±.02	.000001
Q (lpm)	low Psg	8	69.25±16.1	33.16±17	.002
	high Psg	8	89.7±22.4	50±22.4	.007
	combined	16	79.5±21.6	41.6±21.1	.00002
CPP (dB)	low Psg	8	10.96±2.3	14.1±.1.8	.01
	high Psg	8	11.34±2.4	14.11±.1.7	.03
	combined	16	11.3±2.3	14.34±1.7	.0005

The fact that AA with TT1 had a greater effect on VE at higher Psg values is further illustrated by plotting VE over a range of Psg (Figure 2). This additional phonation data over a range of Psg was captured in just one of the larynges (L5) by continuously increasing Psg at 2 cmH2O increments from its PTP level until the vocal fold vibration became irregular. These data show that all experimental states have similar VE at the lowest level of aerodynamic power (defined as Psg multiplied by Q). This includes the UVFP experimental state (i.e., before the TT1 procedure). However, as Psg is increased, the VE for TT1 with AA increases at a similar rate to baseline (i.e., normal phonation), while the VE for TT1 increases at a lower rate.

Figure 2-.

VE plotted against the aerodynamic power for all experimental states in one larynx (L5). As Psg increases, VE increases similarly to normal phonation, while TT1 alone increases at a lower rate closer to UVFP (i.e., without medializing the paralyzed fold).

The intraglottal velocity fields are shown in one larynx (L6) for the closing phases when the highest glottal circulation is computed (Figure 3). The rainbow colors indicate contours levels based on the axial component of the velocity, where positive values (warm colors) mark the glottal jet and negative values (cold colors) imply entrainment flow. It shows the difference in vertical asymmetry between TT1 and TT1 with AA. The vertical asymmetry is formed by the superior displacement of the vocal process secondary to the lateralization of the arytenoid. In the example shown, this vertical displacement of the medialized fold is near 1.25 mm, which is more than 40% of the glottal height measured in the baseline case (about 3mm). The addition of an AA stitch lowers the paralyzed fold. Lowering the vertical height also enables more entrainment flow to feed the FSV that develops between the wall and the glottal jet. This is shown qualitatively by plotting the flow’s streamlines and observing that larger vortices are formed in TT1 with AA. The circulation strength of these vortices also increases correspondently (Table 1).

Figure 3-.

Intraglottal velocity fields in the same larynx (L6). Low and high Psg are shown in the left and right columns, respectively. TT1 and TT1 with AA are shown in the upper and lower rows, respectively. The difference in glottal asymmetry and FSV between the surgical procedures can be appreciated.

DISCUSSION

The current study shows a statistically significant change in VE and CPP when AA is added to TT1 in an excised canine larynx model. The study also shows a decrease in the mean phonatory flow rate (Q) and an increase in the FSV strength with AA. These results agree with previous clinical studies that showed the maximum phonation time increases and phonatory flow rate decreases in patients with AA.^16,34–36 The improvements in these acoustic and aerodynamic features support the paradigm that adding AA to TT1 can result in better surgical outcomes when compared to TT1 alone.

Change in these acoustic and aerodynamic parameters stems from alteration to the glottal geometry when AA is included. AA reduces the glottal asymmetry formed when the implant is inserted through the thyroplasty window in the thyroid cartilage ipsilateral to the paralyzed fold. This asymmetry can be appreciated in CT scans taken in one of the excised larynges following the experimental work described above (Figure 4). The vertical asymmetry is evident in the coronal image for TT1 alone (Fig. 4a). This asymmetry is mitigated when the AA is added to the paralyzed fold (Fig. 4b).

Figure 4 –

CT scans taken in one of the excised larynges (L3) after collecting experimental data. a) TT1 b) TT1 with AA.

Glottal symmetry during phonation supports the formation of stronger FSV. Past studies have shown that FSV strength is correlated with increased acoustic intensity, increased intraglottal negative pressure, and faster closing of the folds during vibration (i.e., higher MFDR).^37,38 It was also shown that their strength is diminished when strong asymmetries are present in the glottis.²⁰ Because AA reduces the glottal asymmetry, stronger FSV are formed compared to TT1 alone.

The increase in VE occurs because changes to the glottal geometry also affect the glottal flow. In all cases, VE increases when AA is added because it closes the posterior gap between the folds. This closure results in the folds vibrating with less glottal flow (Q) when compared to vibrations at the same Psg with a larger posterior glottal gap (i.e., TT1 alone). The lower values of Q can partially explain the improvement in VE values, but it does not explain why the trend in significance level was different between Q and VE at high and low Psg. Specifically, the change in VE was more significant at high Psg (P=0.00002 compared with P=0.004), while the change in Q was more significant at low Psg (P=0.002 compared with P=0.007). This difference, however, could be related to the effect of FSV. We observed that at low Psg, when FSV is smaller, the difference in circulation strength, and thus its impact on phonation, is less pronounced than at high Psg. This association between greater VE and stronger FSV was shown in a previous study using the excised canine larynx model.²⁹

Clinically, increased VE translates to easier, more sustainable phonation. This reduced efficiency causes increased vocal fatigue, a common complaint of UVFP patients. Our results suggest that height differences between the paralyzed and the non-paralyzed vocal fold and the size of the posterior gap need to be considered in surgical planning. It is impossible to say whether the benefit shown for adding AA to TT1 was related to the larger posterior gap that naturally occurs in the canine larynx. These findings will need to be confirmed in a future clinical study. The glottal gap size was considered categorical (i.e., large/small), and therefore we cannot make any recommendations. However, combining the current results with previous work with excised larynges and the study by Wong et al.³⁹, which found that the revision rate of a TT1 was higher for patients with a height asymmetry, suggests that AA could benefit patients with vertical asymmetry > 1 mm after medialization.

Although it suggests that TT1 with AA could benefit UVFP patients with a larger posterior glottal gap or those who need to improve their vocal efficiency, it is not appropriate for all patients. Not all patients can bear the longer operative time required to perform AA. Additionally, AA has a higher surgical risk in terms of swelling, bleeding, dysphagia, and hypopharyngeal lesion. Assessing the need for AA can be done intraoperatively by gauging the vertical asymmetry after medialization and/or measuring the patient’s VE. The latter can be done using the Phonatory Aerodynamic System (i.e., PAS, Pentax Medical) or the Aeroview System (i.e., Rothenberg mask, Glottal Enterprises).

Limitations.

The study’s main limitation is that the excised canine larynx model does not consider supraglottic compensatory mechanisms. These mechanisms are more likely to occur in UVFP patients with small gaps. Our study also did not address differences in breath force. It is known that patients with a poor pulmonary reserve cannot increase subglottal pressure and therefore increasing their vocal efficiency is important.

CONCLUSION

A better understanding of how thyroplasty procedures alter the aerodynamics of phonation will help inform surgical decisions. Our findings suggest a clear acoustic and aerodynamic benefit to the addition of AA when performing TT1. It shows significant improvement in CPP, translating to decreased breathiness and dysphonia, and increased VE, leading to easier and more sustainable phonation. These findings could benefit UVFP patients with vocal height asymmetry after medialization, but more clinical studies should be done to determine appropriate consequences.