Effect of Superior Laryngeal Nerve Stimulation on Phonation in an in Vivo Canine Model

Home , Glottis

Am J Otolaryngol 10:181-187, 1989 Effect of Superior Laryngeal Nerve Stimulation on Phonation in an In Vivo Canine Model

GERALD S. BERKE,MD, DENNIS M. MOORE,MD, BRUCER. GERRATT, PHD, DAVID G. HANSON, MD, AND MANUEL NATIVIDAD

We investigated the effect of variation in superior laryngeal nerve stimulation (SLNS) on vocal fold vibration. Photoglottography (PGG), electroglottography (EGG), and subglottic pressure (Psub) were measured in seven mongrel dogs using an in vivo canine model of phonation. The PGG, EGG, and Psub signals were examined at three SLNS frequencies (100 Hz, 130 Hz, and 160 Hz) using a constant rate of air flow. Increasing SLNS, which causes a contraction of the cricothyroid muscle, produced a marked increase in F,, little change in Psub, an increase in the open quotient, and a decrease in the closed quotient of the glottal cycle. AM J OTOLARYNGOL 10:181-187. 0 1989 by W.B. Saunders Company.

Although theoretical models have been impor- termed “the external tensor” of the vocal cord. tant in providing information about phonatory Through action potentials carried in the external control mechanisms, verification of current branch of the superior laryngeal nerve (SLN), CT theories will require the use of physiologic prep- activation produces a lengthening and thinning arations. This investigation was designed to ex- of the cords. Human studies using high-speed amine the effect of changes in vocal fold mass laryngeal photography during phonation have and tension on vocal fold vibration, using an in demonstrated a lengthening and thinning of the vivo canine model of phonation. The effect of cords with rising Fo,l as well as an increase in different levels of superior laryngeal nerve stim- electromyographic activity of the CT muscle.2*3-5 ulation (SLNS) and recurrent laryngeal nerve It has been shown that stimulation of the SLN in stimulation (RLNS) under conditions of constant dogs produced F, increases from 135 Hz to 540 air flow was studied photoglottographically and Hz.~ Lengthening and thinning the vocal folds electroglottographically, and subglottic pressure affects F, by decreasing the effective vibrating (Psub) was measured. This article reports the re- mass. In addition, lengthening the vocal cord sults of SLNS. causes elongation of the vocalis muscle (VOC). It has also been shown that passive and active ten- LARYNGEAL CONTROL FACTORS sion in the VOC were greatly enhanced by external stretchinga CT activation then elevates F, by The fundamental frequency (F,) of phonation lengthening the VOC, thereby increasing the depends on the effective mass and stiffness of stiffness and reducing the mass per unit area of the vocal folds interacting with transglottal pres- the vocal folds. sure. The cricothyroid muscle (CT) has been IN VW0 CANINE MODEL

Received November 6, 1988, from the UCLA Division of The dog has been the principal animal model Head and Neck Surgerv.Wadsworth VA Larvnaeal Phvsiol- ogy Laboratory, UCLA School of Medicine, L&Angeles. Ac- for laryngeal studies. The canine larynx is simi- cepted for publication November 23, 1988. lar to the human larynx in size and vocal fold Supported by a Veterans Administration National Merit histology; however, the upper portion of the vo- Review grant and a grant from the National Institutes of Health. cal fold has a thicker and looser lamina propria Address correspondence and reprint requests to Gerald S. than humans, resulting in increased thickness of Berke, MD, Laryngeal Physiology-Laboratory, West Los An- the vocal fold.’ The canine larynx also has a geles VA Medical Center, Los Angeles CA 90073. Q 1989 by W.B. Saunders Company. postglottic space in some animals, and, during 0196-0709/89/1003-0003$5.00/O phonation, there is a posterior V-shaped chink

181 SUPERIOR LARYNGEAL NERVE STIMULATION

behind the arytenoids. Anastomotic fibers run- lower margin closing.14 Simultaneous monitor- ning between the superior and recurrent laryn- ing of PGG and EGG signals provides informa- geal nerves in the dog (Galen’s nerve) are be- tion for peak glottal opening and glottal closure lieved to be sensory in nature. Longitudinal elas- similar to high-speed laryngeal photography.15 ticity curves of the epithelium, ligament, and muscle of the canine larynx have different ten- Experimental Preparation sion-length slopes than those for humans, but their overall shape is similar. The cricoid and Our experimental setup was described in a thyroid cartilages are more angulated and previous study and is similar to that of prior in shorter in dogs; the ventricles are considerably vivo canine studies.l”s6 Each dog was anesthe- larger and the vocal ligament is not well-defined. tized with 2 mL ketamine by intramuscular in- In spite of these differences, much information jection followed by intravenous pentobarbital concerning the mechanics of vocal fold vibration until loss of the cornea1 reflex was achieved. The has been derived from studies of excised canine animal was then placed supine on an operating larynges. table (Fig 1) and direct laryngoscopy was per- Some studies have suggested that excised la- formed to confirm normal laryngeal anatomy. A rynges do not reproduce physiologic conditions T-mm oral endotracheal tube was inserted, of vocal fold tension and mass during vibration through which the animal breathed spontane- with sufficient accuracy.g-ll The in vivo canine ously. A midline incision was made from the model appears to be a more physiologically valid mandible to the sternum. The strap and sterno- preparation for studying vocal fold vibration cleidomastoid muscles were retracted laterally than the excised larynges because blood flow to expose the larynx and trachea. The external and intrinsic laryngeal muscular tension are branch of the superior laryngeal nerves were iso- maintained while preventing postmortem tissue lated at their entrance into the CT muscle. A degeneration. These are critically important fac- gauze/silver electrode was applied to the nerves tors to consider when applying the results from and insulated from the surrounding tissue. The models of vocal fold vibration to human phona- recurrent laryngeal nerves were isolated 5 cm tion. inferior to the larynx. Electrodes were applied in the same fashion. Ground electrodes were su- METHODS tured to the trachea and connected to the anode of the nerve stimulator. Electrical isolation be- Subjects tween RLNS and SLNS was verified by direct observation. Maximal stimulation of the recur- Seven adult male mongrel dogs (weighing 25 rent laryngeal nerves, to the point at which the to 30 kg) were selected. Each dog was screened strap muscles were noted to contract (approxi- by direct laryngoscopy to assess its suitability as mately 9 volts), was not observed to produce a subject for the experiment. Dogs with long contraction of the cricothyroid muscle. In addi- necks were preferred for ease of preparation. tion, no lengthening or thinning of the vocal cords occurred during maximal RLNS. Isolated Glottographic Techniques maximal stimulation of the superior laryngeal nerves to the point at which the strap muscles Glottography has proven useful in the study of were observed to contract did not demonstrate temporal events during vocal fold movement.l’ tensing or bulging of the vocalis muscle on direct Photoglottography (PGG) uses a photoelectric laryngoscopic observation. No arytenoid adduc- transducer to measure transillumination of light tion or phonation could be elicited by maximal through the glottis during phonation.13 As the SLNS. EGG electrodes (Synchrovoice, Briarcliff vocal folds vibrate, the intensity of light trans- Manor, NY) were placed in direct contact with mitted through the glottis reflects the cross- the thyroid cartilage while the reference elec- sectional area of the glottis over time. Electro- trode was sutured to the skin. A l.O-cm button glottography (EGG) is a technique measuring the was placed to suspend the epiglottis anteriorly impedance of a small electric current across the through the thyrohyoid membrane to improve neck in the vicinity of the vocal folds. Changes in visualization of the vocal folds. A distal tracheotomy was performed and an endotracheal tube American impedance are modulated by changes in lateral passed to permit the animal to breathe sponta- Journal vocal fold contact area, and the differentiated of EGG signal (dEGG) can provide temporal infor- neously. A more proximal tracheotomy was performed, through which a cuffed tracheotomy Otolaryngology mation on points of upper margin opening and

182 BERKE ET AL

Figure 1. Diagrammatic representation of experimental preparation. (Reprinted with permission.17)

3* PHOTOSENSOR

Recurrent laryngeal n.

tube was placed with its tip resting 10 cm below tion, an air flow of approximately 375 cc/set is the glottis. A catheter-tipped pressure trans- required to develop a Psub of at least 20 cm H,O, ducer was inserted through this upper tracheot- and to match target frequencies of from 80 to 160 omy. The cuff on the superiorly directed tube Hz. The air was bubbled through 5 cm H,O for was inflated to just seal the trachea. Air flow, warming and humidification, and the tempera- obtained from the UCLA physical plant, was ture in the animal’s trachea was measured at 15- passed through the cephalad tracheotomy tube. minute intervals to assure a constant air flow The rate of air flow was measured with a flow- temperature of 37°C. The PGG light sensor (Cen- meter (Gilmont Instruments, model F1500, Great tronics OSD 50-2, Mountainside, NJ) was placed Neck, NY) and kept at a constant rate of 375 cc/ on the animal’s trachea approximately 3 cm be- set throughout the study. Unlike human phona- low the larynx. A xenon light source and fiberop- tion, which can be induced with Psub in the tic cable provided supraglottic illumination for range of 6 to 10 cm H,O, canine phonation re- the PGG. A microphone (Sennheiser, Culver quires at least 20 cm of water pressure for sus- City, CA) was placed 15 cm from the vocal folds tained oscillation. Because of the canine’s pos- and connected to a Storz model 8000 laryngo- Volume 10 terior glottic chink, which allows a significant stroboscope (Culver City, CA) for frequency anal- Number 3 DC (constant) escape of air flow during phona- ysis of the phonatory sound. In addition, strobo- May 1989

183 SUPERIOR LARYNGEAL NERVE STIMULATION

scopic video imaging was obtained using the stimulation were performed to achieve the five Storz stroboscope unit connected to a Storz 0 target frequencies, which were obtained in a ran- degree telescope via a fluid-filled light cable. dom order in all subjects. At 80 and 180 Hz, The xenon light source for the PGG was con- phonation could not be achieved in some ani- nected to the other light port of the telescope. mals, so statistical analysis was limited to the The image from the 0 degree scope was recorded middle target frequencies of 100, 130, and 160 by a Circon CCD video camera and a Sony Hz. &inch video tape recorder (Teaneck, NJ). Al- though a low level of constant xenon light Data Analysis source was present during stroboscopic video re- cording, excellent stroboscopic video imaging The recorded PGG, EGG, and Psub waveforms was obtained.*’ The system was not used for ob- were low pass filtered at a corner frequency of jective measures; however, it was useful for the 1,500 Hz and digitized at a rate of 20 kHz. A interpretation of vibratory events recorded con- 0.5~second sample of stable phonation was used currently with the PGG and EGG signals. in the analysis. Figure 2 shows a representative A catheter-tipped pressure transducer (Medi- glottic cycle. Moments of glottal opening and cal Instruments DCE-1, Hackensack, NJ) was cal- closing were the same as those that have been ibrated at 37°C by submerging it in a water bath reported previously. 14-16Point Ai marks the ini- in 37°C to a depth just covering the sensor (0.5 tial separation of the lower vocal fold margins, cm). The catheter was then calibrated against a Hg manometer from 0 to 120 cm HZ0 pressure. Ai Bi Ci Di *i+l Stimulus

A Grass (model 54H; Quincy, MA) nerve stimulator was used to provide variable voltage stimulation, while a WPI (301-T; New Haven, CT) nerve stimulator was used to provide a low level of constant current stimulus. Voltages ranged from 0.5 to 0.9 V for the Grass stimulator, and currents ranged from 0.1 to 0.15 mA for the WPI stimulator. The frequency of stimulus was 80 Hz, with a pulse duration of 1.5 ms for both units.

Data Acquisition

PGG, EGG, and Psub signals were simulta- neously recorded on a four-channel Tannberg FM tape recorder (model l15D; Armonk, NY). The signals were also monitored on two oscillo- scopes (Tektronix 5116, Beaverton, OR; and Hi- tachi V1050-F, Carson, CA) to assess the ade- quacy of the glottographic signals.

Experimental Design I I I 5 10 I I MSEC Seven animals were stimulated to phonate at C DURATION OFVIBRATORY -D target frequencies of 80, 100, 130, 160, and 180 I CYCLE I

Hz, while maintaining a constant air flow of 375 CLOSED cc/set. While delivering a low level of constant Io;“:;;F$/ I 11 I

current stimulus to the recurrent laryngeal UPPER LOWER nerves (0.10 mA), voltage stimuli to the SLNs MARGIN MARGIN American OPENING CLOSING were varied to produce phonation at the target Journal Figure 2. PGG, EGG, dEGG, and Psub signals recorded from of frequencies. This was done to examine the effect a canine preparation phonating in the modal register. (Re- of CT muscle activation on F,. Two trials of SLN printed with permission.“) Otolaryngology

184 BERKEETAL

determined by the initial rise in the EGG imped- 0.6 - ance from its minimum. Point Bi represents the 0.5 .. moment of upper margin opening as determined I/i----_,kQp g 0.4 -- by the positive deflection of the dEGG wave- Lu form. Point Ci identifies the moment of maximal F 0.3 -- glottal area as determined by the peak of the PGG 0 2 0.2-. waveform. Point Di marks lower margin closure, as determined by the lowest point in the dEGG 0.1-- 6 6 6 Qcg waveform. Point Ai + 1 is the lower margin open- 0 -I , ing of the next cycle. Points Bi, Ci, and Di could 100 130 160 be reliably determined; however, point Ai, dur- FREQUENCY(Hz) ing high SLNS, occurred on a gradual increase in Figure 3. Mean values (*SEM) for Qp, OQ, Qog, CQ, and the EGG waveform and was difficult to deter- Qcg for seven subjects for SLNS at 3 TF, (100, 130, and 160 mine. These periods occurring within the glottal Hz). (Reprinted with permission.“) cycle were divided by the total period of the vibratory cycle to determine the quotients of vocal fold vibration as described in the Appendix. for SLNS. With ascending F,, the Qog increased Ten contiguous glottal cycles were analyzed [F(2,12) = 4.74; P < .05] in parallel with the and the mean of these cycles was calculated. open quotient (O(z), whereas the closed quotient This procedure yielded means of glottal quo- (Cq) decreased markedly [F(2,12) = 4.50; P < tients as the five target frequencies for two ran- .05]. Newman-Keul’s post-hoc multiple compar- dom trials. Measured F, was also compared to isons revealed significant differences among the target F,. Mean Psub maximum/minimum and three F, levels for these three quotients. Neither root mean squared were obtained for each glottic Qcg, SQ, nor Qp were shown to change signifi- cycle analyzed. cantly at P = .O5 as F, increased. The six glottal quotients were analyzed using Figure 4 shows representative waveforms plot- analysis of variance (ANOVA). Separate analy- ted for one subject at three target frequencies. As ses of variance were applied for each quotient the frequency increased with the increasing with trial as a repeated measure. Each trial was a level of SLNS (Fig 4), the width (Bi - Di) of the set of recordings for each five target frequencies. EGG waveform remained nearly constant, de- No significant difference was shown between tri- spite a diminishing period. This was reflected by als 1 and 2 at P = .05 for any of the quotients, so an increasing open portion (Bi - Ci) and a de- both trials were combined for another set of creasing closed portion (Ci - Di) of the cycle. In ANOVAs with F, as a repeated measure. this particular subject, mean Psub increased from 22 cm H,O at 100 Hz to only 32 cm H,O at RESULTS 160 Hz for SLNS (Fig 4).

Table 1 displays mean F, values obtained by DISCUSSION SLNS for the seven subjects’ target frequencies. The experimentally measured F,s closely ap- This investigation studied the effect of SLNS proximated the desired target frequencies. on physiologic events within the vocal vibratory Figure 3 displays temporal events in the glot- cycle over a frequency range of 80 to 180 Hz tal cycle as glottal quotients by target frequency under conditions of constant air flow. This range was chosen to represent frequencies within the TABLE 1. Target Frequency Versus Measured canine modal register. In a previous study of Frequency for Seven Subjects* adult human males, a range of 94 to 287 Hz was found for the modal register of 12 subjects.lg The SLNS dog likely has a lower range of frequency for the Target F, (Hz] Measured F, (Hz) modal register than the human, because the ca- 80 80.2 (1.1) nine larynx is somewhat thicker and more 100 101.8(1.7) massive.” 130 130.1(4.3) 160 157.1(3.7) The profound capacity of SLNS to increase F, 180 178.8(4.2) was confirmed by this study. Frequencies as high as 340 Hz were recorded during activation Value is mean of two trials for seven subjects over ten Volume 10 * of the CT muscle. F, elevation by increasing contiguous glottal cycles. Numbers in parentheses are stan- Number 3 dard deviation. SLNS was not accompanied by a significant in- May 1989

185 SUPERIOR LARYNGEAL NERVE STIMULATION

offset by a decline at 160 Hz so that no net change in Qp was observed. The difficulty in determining the point Ai with increasing SLNS may have contributed to an error in the calcula- tion of Qp at 160 Hz, thus leading to the observation of no net change. The close association between Qog and OQ for SLNS (Fig 4) was another interesting finding. Most of the change in OQ was the result of upper margin opening, rather than of lower margin 5 10 15 20 closing (Qcg). As F, increases with increased CT MS activity, the opening phase of the vibratory cycle increases while the closing phase remains un- changed. This substantiates the hypothesis that as CT activity is increased in the modal register, a more convergent glottis is produced such that the upper margin governs changes in the open portion of the cycle. Stroboscopic video imaging during increasing SLNS revealed reduced lateral excursion of the 30 0, vocal cords, but this was associated with an ap- 25 = 5 parent reduction in the amplitude of the travel- 20 ing wave. As SLNS was increased beyond modal 5 IO 15 20 voice, the two-margin system was replaced by a MS one-margin system; that is, the lower and upper margins were observed to fuse into a vibration of one mass. Both RLNS and SLNS demonstrated the ability of the CT muscle to thin and stretch the folds for any given level of RLNS. As shown in Fig 3, SLNS produced little increase in Psub with increased F,. Considering that air flow was constant, CT activation had little effect on glottal resistance. Our observation that the alternating current pressure waveform is near its maximum before 5 10 15 20 upper margin opening and falls to its minimum

MS at closure does not agree with human data ob- Figure 4. Typical recordings from a subject for three levels tained with pressure-tip transducers passed of target frequency (TF,). A, B, and C have TF, of 100, 130, through the glottis into the trachea.23*24 In those and 160 Hz for SLNS, respectively. (Reprinted with recordings, a great increase in Psub occurs at the permission.“] onset of glottal closure and a marked decrease occurs at the beginning of glottal opening. These crease in Psub, indicating that SLNS has little discrepancies are attributable to a low-pass filter effect on laryngeal resistance. Increasing OQ (300 Hz) in the pressure transducer amplifier with increased F, under SLNS agrees with stud- used in this study, which had the effect of de- ies of increased F, in human speech.‘l This may laying the actual pressure signal by several mil- be explained by the two-mass model with in- liseconds. creased coupling between the lower and upper margins of the vocal cords. The thinning of the CONCLUSIONS vocal cords produces a shorter vertical distance between the lower and upper margins and (1) Increasing SLNS activated the cricothyroid causes a more convergent glottis.” Conversely, muscle, causing a marked increase in F, with an increase in OQ and a small increase in Psub. The American CQ decreased markedly as F, was increased by opening segment of the open phase was most Journal SLNS. This may also be explained by increased of coupling between the lower and upper margins. affected, as evidenced by a significant increase in Qog. Otolaryngology An initial rise in Qp with increased SLNS was BERKE ET AL

(2) CQ decreased markedly with increasing 7. Hast MH: Physiological mechanisms of phonation; tension of the vocal fold muscle. Acta Otolaryngol 1961; SLNS. 62:309-310 (3) Qp was not shown to change significantly 8. Hirano M: Structure of the vocal fold in normal and dis- with increasing SLNS. ease states. Anatomical and physical studies. ASHA Reports 1981; ll:ll-30 (4) Speed quotient was not shown to be signif- 9. Mueller E: Stimmphysiologische Untersuchungen an icantly altered by F, elevation. einem Kehlkopfmodell. Archiv fuer Sprach- und Stimmphysiologie 1938; 2:197-214 Acknowledgment. The authors would like to 10. Perlman AL, Titze I: Measurement of viscoelastic prop- thank Drs Vicente Honrubia and Judy Dubno for the erties in live tissue, in Titze IR, Scherer RC (eds): Vo- use of their technical support and laboratories. The cal Fold Physiology. Denver, CO, Denver Center for authors are indebted to Dr Ted Bell for his assistance the Performing Arts, 1983, pp 271-281 with statistical analysis and experimental design. Fi- 11. Fukuda H. Saito S. Kitihara S. et al: Vocal fold vibration nally, the authors wish to express their gratitude to Dr in excised larynges: View with an x-ray stroboscope and an ultra-high-speed camera, in Bless DM, Abbs JH Larry Hoffman for his expertise in neurophysiology. (edsf: Vocal Fold Physiology. San Diego, College Hill Press, 1983, pp 238-252 APPENDIX A. Formulas Used in 12. Gerratt BR, Hanson DG, Berke GS: Glottographic mea- Glottographic Measurement sures of laryngeal function in individuals with abnor- mal motor control. in Harris K. Sasaki C. Baer T feds): Qp = time delay between opening of lower and upper mar- Vocal Fold Physiology: Laryngeal Function in Phona- gins/period of glottal cycle: tion and Respiration. San Diego, College Hill Press, Bi - Ai/(Ai+ I) - A, 1987, pp 521-532 13. Sonesson B: A method for studying the vibratory move- OQ = duration of open glottis/period of glottal cycle: ments of the vocal cords. A preliminary report. J Di - Bi/(Ai+,) - Ai Laryngol Otol 1959; 73:732-737 14. Childers DG, Krishnamurthy AK: A critical review of CQ = duration of complete glottal closure/period of glottal electroglottography. CRC Crit Rev Biomed Eng 1985; cycle: 12:131-161 (A,+11 - Di/(Ai+,) - Ai 15. Baer T, Lofquist A, McGarr N: Laryngeal vibrations: A comparison between high speed filming and glotto- Qog = duration of glottal opening/period of glottal cycle: graphic techniques. J Acoust Sot Am 1983; 73:1304- Ci - Bi/(Aj+,) - Ai 1307 16. Berke GS, Hantke DR, Hanson DG, et al: An experimental Qcg = duration of glottal closing/period of glottal cycle: model to test the effect of change in tension and mass Di - C,/(A,+,) - A, on laryngeal vibration, in Hirano M (ed): Proceedings of the International Conference on Voice. Kurume, Ja- SQ = duration of glottal opening/duration of glottal closing: pan, Kurume University Press, 1986, on l-8 Ci - B,lDi - Ci 17. Moore DM, Berke GS: J Acoust Sot Am 1988; 83:705-715 18. Berke GS, Moore DM, Hanson DG, et al: Laryngeal mod- eling: Theoretical, in vitro, in vivo. Larvnaoscope- ., 1987; 97:871-881 References 19. Hollien H, Michel JF: Vocal fry as a phonational register. J Speech Hear Res 1968; 11:600-604 1. Moore P, von Leden H: Sound Film: The Larynx and 20. Hirano M: Morphological structure of the vocal cord as a vibrator and its variations. Folia Phoniatr 1974; 26:89- Voice--The Function of the Normal Larynx. 1956 94 2. Arnold GE: Physiology and pathology of the cricothyroid 21. Hildebrand BH: Vibratory patterns of the human vocal muscle. Laryngoscope 1961; 71:687-753 cords during variations in frequency and intensity, 3. Faaborg-Anderson K: Electromyographic investigation of PhD dissertation, University of Florida, 1976 intrinsic laryngeal muscles in humans. Acta Physiol 22. Ishizaka K, Flanagan JL: Synthesis of voiced sounds from Stand 1957; 41:140 (suppl) a two mass model of the vocal cords. Bell System Tech 4. Hirano M, Ohala J, Vennard W: The function of laryngeal J 1972; 51:1233-1268 muscles in regulating fundamental frequency and in- 23. Kitzing P, Lofqvist A: Subglottal and oral pressure during tensity of phonation. J Speech Hear Res 1969; 12:6X6- phonation-Preliminary investigation using a minia- 628 ture transducer system. Med Biol Eng 1975; 13:644- 5. Shipp T, McGlone RE: Laryngeal dynamics associated 648 with voice frequency change. J Speech Hear Res 1971; 24. Cranen B, Boves L: Pressure measurements during 14:761-768 speech production using semiconductor miniature 6. Rubin HJ: Experimental studies on vocal pitch and in- pressure transducers: Impact on models for speech tensity in phonation. Laryngoscope 1963; 73:973-1015 production. J Acoust Sot Am 1985; 77:1543-1551

Volume 10 Number 3 May 1989

187