Recurrent laryngeal nerve paralysis: Current concepts and treatment: Part II--Causes, diagnosis, and management.

[Editor's note: This article, reprinted from Annales d'Otolaryngologie et de Chirurgie Cervico-Faciale [*], is Part II in a three-part series. Part I-- "Phylogenesis and physiology," appeared in the November 2000 issue of Ear, Nose & Throat Journal and included the references for the entire three-part article. Part III--"Surgical options" will appear in a future issue of Ear, Nose & Throat Journal.]

Causes of URLNP

The exact incidence of unilateral recurrent laryngeal nerve paralysis (URLNP) is unknown, [41] although it seems to increase with age, due to the age-related increase in neurologic and neoplastic pathology that can cause URLNP. Any unilateral damage in the central or peripheral nervous system involving the motor neurons or the axons of the RLN can cause URLNP.

Surgical trauma seems to be the principal etiology, with thyroidectomy being the most frequent intervention causing URLNP. The relative frequency of the different causes seems to have changed during the past 30 years. According to Benninger et al, the incidence of URLNP after thyroidectomy has decreased, while for other interventions (cervical spine and carotid surgery, especially) the incidence is increasing. [41] Skull base surgery and thoracic surgery may also cause URLNP. Car and motorcycle accidents and intubation trauma are the most frequent nonsurgical causes of URLNP.

Extralaryngeal neoplasm, especially pulmonary carcinoma, seems to be an increasing cause. Aortic aneurysm, cardiomegaly, and tuberculous sequelae (to name a few) may cause compression of the vagus nerve or the RLN. Inflammatory neuritis (by the varicella zoster virus, for example) has been reported, as well as toxic neuritis (vincristine), but these etiologies are relatively rare.

Ten to 27% of cases of URLNP are idiophathic. [12] Published series differ widely in etiologic distribution, probably due in part to a selection bias according to the subspecialization of the department (a trauma center vs. an oncology center, for example).

The incidence of "idiopathic" URLNP would also depend on the extent of the patient's workup. An overestimation of idiopathic URLNP may have occurred before routine use of computed tomography and magnetic resonance imaging. However, an underestimation may persist in that only symptomatic URLNP is seen by a physician. A certain number of well-tolerated idiopathic URLNP with slight symptoms and favorable outcome may be underdiagnosed as laryngitis.

Physical evaluation

Symptoms. Voice is weak and phonation demands an unusual amount of effort. The aerodynamic turbulence of the glottic air leak is perceived as "breathiness." Voice may also have a bitonality or diplophonia, with the perception of two different fundamental frequencies (or pitches) simultaneously. URLNP may cause aspiration, mainly of liquids, which occurs during the pharyngeal phase of swallowing. [42] Postswallow aspiration is also possible and is due to laryngeal penetration of residual food bolus not cleared from the piriform sinus on the paralyzed side. Inspiratory dyspnea or stridor may occur, most often noticed during physical effort. [43] Cough is less efficient in URLNP. And some patients complain of paresthesia or a foreign body sensation in the larynx. These symptoms are generally more pronounced in cases of vagus nerve damage as opposed to pure RLN damage, but interpatient variability is such that the lesion site or cause cannot be determined based only on symptoms. [44]

Physical signs. Laryngeal configuration and mobility are best evaluated using fiberoptic laryngoscopy or rigid-tube laryngoscopy, which provide enlargement and recording of the image for further study and evaluation of evolution. Laryngeal stroboscopy is recommended for a detailed evaluation of vocal fold vibration and is especially useful for postoperative evaluation.

All the intrinsic laryngeal muscles are affected by URLNP except for the IA, which is bilaterally innervated, and the CT, which is innervated by the external branch of the SLN. A vagal nerve lesion will also affect the CT. The vocal fold and arytenoid are immobile on the paralyzed side, resulting in vibration asymmetry and a glottal gap on phonation. Both the asymmetry and the glottic incompetence are responsible for the abnormal acoustic quality and the decreased aerodynamic efficiency of the larynx.

Glottic configuration at rest. The paralyzed vocal cord (VC) has a variable position in the axial and frontal planes. The "traditional" VC positions are paramedian (0 to 1.5 mm from the midline) or intermediate (1.5 to 2.5 mm from the midline) for isolated RLN lesions. The intermediate or lateral position ([greater than or equal to]2.5 mm from the midline) is said to occur more frequently in vagus nerve lesions. Due to the large variability between patients, however, this difference is not reliable in determining the origin of the paralysis. [38]

VC position in the frontal plane is also variable. The paralyzed VC can be at the same level as the contralateral VC, lower or higher according to the position of the arytenoid. The paralyzed VC is lower if the paralyzed arytenoid falls forward and medially. [26,45] A higher, more cranial position of the VC results if the vocal process of the arytenoid is displaced in a posterosuperolateral direction." This vertical asymmetry should be carefully noted, as it determines the surgical indication, especially that of arytenoid adduction, which medializes and lowers the vocal process. [44]

The paralyzed VC may be bowing and flaccid at rest from loss of muscle tone and from loss of tension due to the anterior displacement of the arytenoid. The arytenoid on the paralyzed side may be in a neutral position or, as stated above, may fall forward and medially, often with a passive falling of the corniculate cartilage, which can impede view of the posterior glottis.

In some cases, laryngeal rotation occurs. It is evaluated by determining the axis between the anterior and posterior commissures as compared with the frontal plane, but it is relatively difficult to measure in a reproducible manner. [46,47] Rotation is inconstant, but it seems to be more frequent and pronounced in vagal lesions as compared with RLN lesions. [48] Woodson has shown that the CT muscle does not intervene in this laryngeal rotation, [24] but no other explanation has been proffered to explain this phenomenon.

At rest during quiet breathing, a "parasitic" motion of small amplitude can often be observed, especially in the region of the arytenoid. This motion remains unexplained, but it seems to arise at least in part from passive aerodynamic effects on the mucosa. Residual innervation in the IA does not seem to play a role. [38] Accumulation of saliva may be observed in the piriform sinus on the paralyzed side. This may cause aspiration in some patients, but it can be present without aspiration. Swallowing studies using fiberoptic laryngoscopy or radiologic analysis is necessary to determine the mechanism of aspiration. [42]

Glottic configuration in phonation. Hoffman and McCulloch have described two types of glottic incompetence during phonation in URLNP. [45] The first is incomplete closure along the membranous VC only, with an elliptical gap and no posterior glottic gap. This configuration is seen when the VC is in a paramedian position. The second type is a triangular glottic gap involving both the membranous VC and the posterior glottis. This type seems to correspond to an intermediate or abducted VC. This distinction has implications for surgery in that some techniques (vocal fold injection, for example) do not sufficiently treat posterior glottic gaps.

Videostroboscopy provides objective analysis of the phase, the speed, and the amplitude of the mucosal wave. The stroboscopic image is a combination of a large number of instantaneous pictures of the glottis at different moments during the glottic cycle. Detection of fundamental frequency (F0) is necessary to control the timing of the light flashes of the stroboscope. Voicing is sometimes of such poor quality in URLNP that detection of the F0, and thus stroboscopy, is impossible.

VC vibration is defined as the lateral-to-medial motion of the VC edge during phonation. According to the body-cover model described by Hirano, [14] this motion depends on certain characteristics of the vocalis muscle itself. Vibrations are asymmetric in URLNP due to the asymmetry in mass and muscle tone of the two VC. [49] The mucosal wave is defined as the motion of the mucosal cover, which effectuates an elliptical "rolling" motion in the frontal plane, moving from caudal to cranial and from medial to lateral. The mucosal wave depends on certain characteristics of the mucosa and on the state of the superficial lamina propria, essentially a buffer between the body and the cover. Sercarz et al have demonstrated an asymmetry in the mucosal wave, with a delayed initiation, slower period, and lower amplitude on the paralyzed side.48 Glottic insufficiency can be such that the mucosal wave is not visible on the paralyzed side, either because of its absence or because its small amplitude surpasses the limits of re solution of the stroboscope. Absence of a mucosal wave is not proof of severe nerve damage but rather of unfavorable aerodynamic conditions in the paralyzed larynx. Stroboscopy when performed should be carried out for as long a voicing as possible and video-taped if possible, for VC vibration and mucosal wave may vary from one glottic cycle to another.

In summary, there seems to be no significant difference between high vagal lesions and RLN lesions in terms of VC position or length, degree of abduction on inspiration, anterior commissure angle, arytenoid position, glottic configuration on phonation, and vibrational asymmetry. [38,48] A detailed evaluation of the glottic configuration is nonetheless essential in order to choose the most relevant surgical indication. In addition, glottic configuration evolves with nerve regrowth, and modifications may indicate a favorable long-term outcome.

Phonatory compensation. During phonation, the normal VC may adduct beyond the midline and "compensate" for the abducted position of the paralyzed cord. This phenomenon can be seen immediately after the onset of URLNP or it can appear secondarily. This configuration may lead to a favorable functional outcome despite persisting paralysis. The neural and muscular mechanisms are unknown. Voice therapy has never been shown to increase the incidence of this type of compensation.

Supraglottic structures may also intervene to "compensate" during phonation. It is not clear if the supraglottic hyperactivity observed in URLNP is active or passive. The ventricular fold (false cord) on the nonparalyzed side adducts with the normal VC. False cord adduction corresponds to type II supraglottic hyperfunction, according to Koufman and Blalock. [50] Supraglottic compensation with anteroposterior compression of the arytenoids, the aryepiglottic folds, and the epiglottis (type III [50]) is also possible. [38] In some cases, an increase in the tension of the VC by CT contraction is visible and audible, resulting in a "paralytic falsetto." [46] A final mode of compensation is complete supraglottic dysphonia with vibrations originating at the false cords, the arytenoids, and/or the epiglottis (Koufman's type IV [50]).

History of theories on glottic configuration

Galen, by sectioning the RLN in the pig, was the first to prove that voice came from the larynx. Before his 2nd century experiments, voice was thought to arise from the heart. [51] Turck in 1860 was the first to describe and publish a laryngoscopic description of URLNP. [52] In 1863, Gerhardt hypothesized that the different glottic configurations observed were due to lesions at different levels along the nerve, with a bilateral central lesion causing bilateral abduction or the "cadaver" configuration. Fifty years later, Elsberg developed a theory of configuration based on a hypothesized denervation of different combinations of laryngeal muscles. At the same time, Semon and Rosenbach hypothesized that the abductor nerve fibers were more vulnerable to trauma than the adductor fibers due to their relative positions in the RLN, abductor fibers being in a peripheral location and adductor fibers in the center.

Jelenffy and Krause thought that the paramedian position of the VC was the result of external innervation, with an "irritative" stimulation from adjacent peripheral nerves or from the central nervous system. In addition, Jelenffy was the first to describe the antagonistic (abductor-adductor) function of the laryngeal muscles. He described the now classic anteromedial rocking of the arytenoid apex and the inferior shift of the vocal process in URLNP. Later, at the beginning of the 20th century, Solis-Cohen and Casselberry were the first to describe spasmodic contractions and other abnormal motion of the paralyzed larynx. [52]

At the end of the 19th century, the theories of Semon-Rosenbach ("fragile" abductor fibers) and of JelenffyKrause ("irritative" external stimulation) each had firm proponents and opponents. Semon's law prevailed until the 1950s, thanks to a number of experiments conducted by Hooper in the 1880s. Also at the end of the 1900s, Wagner and Grossman explained the variable glottic configurations by variation in CT activity. Their theory was reaffirmed by Dedo in 1970 in several experiments using laryngeal EMG. [52]

From approximately 1950 to 1980, three major theories explained the variability of glottic configuration in URLNP: (1) Wagner-Grossman's theory of CT activity, (2) the existence of extralaryngeal motor branches of the RLN that make injury variable according to the extralaryngeal configuration and the level of the damage, and (3) denervation atrophy and fibrosis of laryngeal muscles. The first and second theories were later refuted, giving way to the current explanation for variable glottic configuration.

Current concepts in glottic configuration

As stated above, Gacek disproved Semon's law by formally demonstrating the random organization of the abductor and adductor fibers within the RLN. [16] Wagner-Grossman's theory was refuted in 1995 by Koufman, who demonstrated the absence of correlation between glottic configuration and EMG activity of the CT. [53] The current model for URLNP is based on the concepts of misdirected nerve regrowth with synkinesis, residual innervation, and muscular atrophy and fibrosis.

Synkinesis. As stated above, synkinesis is the simultaneous nerve-instigated contraction of antagonist muscles. [35,52] The contractions are involuntary and may accompany voluntary contractions. [26] Their neural origin is reflected in the absence of muscular atrophy and the persistence of motor end plates. [40]

Blitzer et al elegantly demonstrated the effect of synkinesis on glottic configuration in URLNP. [26] Two groups of patients having URLNP for at least 1 year were studied. Patients with a subjectively "good" voice had the paralyzed VC in a paramedian position and on the same level as the contralateral VC. The arytenoid at rest was in a neutral position, without anterior or medial rocking. During phonation, some small saccadic arytenoid motion was observed in several patients. For all of these cases, normally shaped motor potentials were detected during phonation, although amplitude was smaller than on the nonparalyzed side. Patients with a poor-quality voice had a paralyzed arytenoid in an abnormal position, with anterior and/or medial shifting. The paralyzed VC was in a paramedian or intermediate position, with some vocal fold bowing. In all cases, motor potentials were absent or abnormal, with fibrillation, low amplitude, polyphasic, or complex shapes.

These findings can be explained by misdirected axonal regrowth, with random assignment of axons in the distal endoneurial tubules, and thus a random reinnervation of abductor and adductor muscles. Synkinesis inhibits muscular atrophy. Reinnervation may remain insufficient to provoke voluntary contraction, but it may be sufficient to hold the arytenoid, vocal process, and VC in a neutral position, favoring good vocal results by contralateral VC compensation. In synkinesis, the vectors of motion during a voluntary command may cancel out each other, resulting in zero net laryngeal motion. This also results in a neutral laryngeal configuration, without VC bowing or atrophy. All of these scenarios result in a good functional result and are thus referred to as "favorable synkinesis." [36] If the motion vectors do not cancel out each other, synkinetic motion may lead to abnormal laryngeal motion with spasmodic dysphonia or dyspnea. [54] This neurologic configuration is referred to as "unfavorable synkinesis." [37]

Laryngeal synkinesis (and laryngospasm) in URLNP can be likened to the synkinesis and spasms occurring after facial paralysis. This classification into favorable and unfavorable synkinesis can be considered as a functional equivalent to House and Brackmann's scale for facial palsy.

Residual innervation. In cases of neurapraxia or axonotmesis (even partial), some axons remain intact with their normal abductor or adductor function. Glottic configuration would thus initially depend on the number and type of residual axons and then on the type of nerve regrowth, with or without synkinesis.

The bilateral innervation of the IA could also contribute to glottic configuration. In addition to pulling the bodies of the arytenoid cartilages together, the IA seems to act as an accessory abductor, effecting a backward (and medial) sliding motion of the arytenoid. Persistent motion of the IA may in part be responsible for a medial shifting of the arytenoid. [37,38] Finally, some of the small-amplitude motion of the paralyzed larynx may in fact be passive, aerodynamically elicited motion. [51]

Denervation atrophy. Denervation atrophy in URLNP has been confirmed by several histopathologic studies. [55-57] Currently, fibrous retraction of the vocalis muscle is thought to be the cause of the intermediate or lateral VC position in completely denervated larynges, occurring after several months. [57] Kirchner [56] in 1966 and Gacek and Gacek in 1996 [58] confirmed this phenomenon and showed the absence of cricoarytenoid joint ankylosis, even in protracted paralysis (17 years in one case). It is possible that passive motion of the paralyzed arytenoid, imposed upon it by the CT, the IA, and/or the external laryngeal muscles, keeps the joint intact. The impression of ankylosis on palpation during endoscopy may be the effect of severe muscular fibrosis surrounding the joint. The absence of true ankylosis permits surgery for arytenoid mobilization and VC medialization, but with protracted paralysis, muscular atrophy and fibrosis may be limiting factors to medialization.

Remaining questions. Where do the misdirected axons come from?

It seems that they arise from the proximal RLN, as demonstrated in the experiments by Flint et al studying RLN section and anastomosis. [40] No reinnervation was found coming from the SLN (which had been cut and not anastomosed) or from the contralateral RLN.

Boles, however, cut the RLN in dogs without anastomosis (the stump was reclined and sutured to the subcutaneous tissue) and still found significant EMG activity 11 months later. [59] This reinnervation could not have arisen from the proximal RLN.

Nomoto et al found positive epinephrine staining in the laryngeal motor end plates following RLN section without anastomosis. [59] They interpreted their finding to imply "colonization" of synapses by the autonomous nervous system. The same colonization by adrenal or cholinergic fibers has also been shown for other skeletal muscles in other animal species. The same authors postulate that these other peripheral nerves produce neurotrophic or musculotrophic factors that prevent muscular atrophy. [60] Thus it is possible that nerve fibers from the autonomous system, normally around laryngeal blood vessels, play a role in reinnervation and prevention of laryngeal atrophy.

Does she internal branch of the SLN contain motor fibers?

As stated above, anastomoses between the SLN and the RLN exist at several points. The external branch of the SLN gives off a branch that runs through the TA and in doing so forms anastomoses with terminal branches of the RLN. Within the IA, a branch of the SLN connects with muscular branches of the RLN. The functional role of these anastomoses, however, is unknown. They may or may not contain motor axons; their neural influx may be efferent or afferent. [25]

Differential diagnosis of URLNP

Mechanical arytenoid immobility is the principal differential diagnosis. Patient history often reveals intubation trauma, external laryngeal trauma, a severe systemic infection, or a chronic inflammatory disease. The immobility may originate in tissues surrounding the cricoarytenoid joint (posterior glottic stenosis) or in the joint itself (arthritis, cricoarytenoid luxation). No VC atrophy is observed, and some limited motion of the arytenoid may be apparent. Differential diagnosis necessitates endoscopic evaluation of arytenoid mobility and/or laryngeal EMG.

False arytenoid immobility may be observed in the case of a supraglottic tumor masking the vision of the arytenoid. The VC remains mobile. VC bowing and atrophy can be seen in elderly patients ("presbylaryngitis") and in cases of sulcus glottidis. The arytenoid has normal mobility, and the sulcus is visible on stroboscopy.

Workup

The diagnostic workup for URLNP of unknown origin includes a complete examination of the head, neck, and upper aerodigestive tract and imaging of the vagus and RLN. Chest x-ray is needed to rule out a mediastinal or pulmonary tumor or other pathology. Most authors recommend computed tomography from the skull base to the diaphragm to evaluate the entire trajectory of the vagus nerve. Other imaging techniques or blood workup are required if the clinical examination leads to a suspicion of a particular disease. If doubt persists as to a mechanical laryngeal immobility, direct laryngoscopy under general anesthesia with palpation of the arytenoid cartilage and evaluation of the posterior glottis is indicated. [51]

Acoustic and aerodynamic evaluation should be performed to evaluate the functional handicap due to URLNP. Subjective acoustic evaluation is based on the patient's and physician's judgement of the dysphonia. Visual or numerical scales are useful for semiquantification of voice quality characteristics. The GRBAS scale is frequently employed. [61,62] G is the degree of global dysphonia, R is roughness, B breathiness, A asthenia or weak voice, and S a strained or forced voice. Generally each parameter is measured on a scale of 0 (absent) to 3 (severe). The visual aspect of the voice signal, the spectrum (after Fourier transform), and the spectrogram are also useful elements in a patient's file, allowing subsequent comparison following spontaneous voice improvement or surgery.

Objective acoustic analysis is meant to quantify dysphonia, but current methods are far from being sensitive and specific enough. Currently available voice software often calculates parameters such as fundamental frequency, jitter (variation of the fundamental frequency), shimmer (variation in amplitude), and the harmonics-to-noise ratio. These parameters give an idea as to the general degree of dysphonia, but they may not be sensitive enough to detect slight improvements in severely dysphonic voices. [5] Other types of objective acoustic analysis are currently being investigated in hopes of improving sensitivity and specificity.

Aerodynamic measurements complement acoustic analysis. One simple measurement especially pertinent in URLNP is that of maximum phonation time--that is, the maximum length of time a patient can phonate a vowel (usually /a/) after maximal inspiration. Other measurements include the number of syllables pronounced per minute or the maximum number of syllables pronounced following a maximal inspiration. These parameters evaluate phonatory air use. They also depend on pulmonary function and general health. Objective aerodynamic measurements are now possible using commercially available flow and pressure transducers, and in the near future they will be a standard part of any voice clinic. [5] Inspiratory breathing difficulties can be objectively measured using pulmonary function tests. The inspiratory volume per second calculated from a flow-volume loop quantifies an eventual laryngeal obstruction during inspiration and permits followup after treatment. [43]

The workup thus includes diagnostic imaging and functional voice and breathing evaluation. No other tests are indispensable. The systematic use of laryngeal EMG is not necessary. It is not always available and it requires a specifically trained neurophysiologist. [5] Results lack reproducibility (even within the same patient), are only qualitative, and do not always provide a definitive diagnosis. EMG is particularly indicated in distinguishing high vagal nerve lesions from RLN lesions based on CT activity. [51] EMG is not considered reliable for predicting prognosis in URLNP. [37,63]

Evolution

An acute phase of several weeks has been described. [37] Symptoms are generally most severe during this phase, with dysphonia or whispered voice, phonatory fatigue, and possibly aspiration. During the intermediate phase, symptoms may improve due to contralateral or supraglottic compensation. For neurapraxic or axonotmetic lesions, signs of laryngeal remobilization appear between 6 and 12 months. [51] In more severe lesions, laryngeal immobility persists with signs of vocalis muscle atrophy. Following the intermediate phase, total recuperation of function, recuperation with synkinesis, or chronic immobility may be observed. [37] It is currently believed that 25 to 60% of patients wit idiopathic URLNP will recover wit normal function, favorable synkinesis, and/or contralateral compensation. [41,64]

Treatment

"Watch and wait." If surgery is not immediately necessary, conservative treatment is a viable therapeutic option. The waiting period after which laryngeal configuration is considered stable depends on the mechanism of the nerve damage. If nerve envelopes have been preserved, functional improvement or synkinesis is likely and generally appears within 12 months. [37] If the nerve has been cut, spontaneous recovery is unlikely without anastomosis. Contralateral compensation can theoretically occur, but this does not seem to be frequently observed. If the mechanism of nerve damage is unknown, regrowth must be hoped for.

During this 12-month waiting period, swallowing must be carefully evaluated to detect eventual asymptomatic aspiration. Elderly patients are especially at risk because pre-existing dysphagia can decompensate with the onset of URLNP and lead to severe aspiration.

Voice therapy. Voice therapy in URLNP remains controversial. [45] Initial evaluation by a speech therapist is considered useful to determine the degree of functional handicap and to detect other voice or speech abnormalities--pre-existing or brought on by URLNP. While awaiting an eventual spontaneous recovery, voice therapy may provide patient information and psychologic support. [37] Information and support are indispensable, and they may also be provided by the otolaryngologist or the family doctor.

Voice therapy has never been shown to improve glottic closure or to speed spontaneous recovery, and in some cases it may worsen voice outcome. [45] Vocal exercises should be directed at developing respiratory support and avoiding excessive tension in laryngeal and cervical muscles.

Manual compression on the thyroid cartilage can be used to momentarily improve voice, increasing vocal intensity and decreasing phonatory effort. Other exercises are useful for developing strength and endurance in respiratory muscles, especially the diaphragm, while preventing excessive tension in cervical and external laryngeal muscles. Controlled, isotonic laryngeal exercises may be useful, but intense, forced movements should be avoided. Exercises that induce forced adduction (lifting and straining) can lead to supraglottic hyperactivity, a type of laryngeal compensation that seems to occur frequently in URLNP. This hyperactivity increases laryngeal resistance and thus decreases phonatory effort, but it causes a rough voice quality. It may be irreversible even after spontaneous recovery of laryngeal motion or after surgery for vocal fold medialization. [45,51] Voice therapy should especially be aimed at preventing this type of compensation when good voice quality is the ultimate goal.

The duration of voice therapy depends on the evolution of voice and laryngeal configuration. Exercises can be continued as long as the patient experiences a benefit.

Laboratory of voice, biomaterials and cervicofacial oncology, CNRSUPRESA 7018, University of Paris V, Laennec Hospital, 42 rue de Sevres, 75007 Paris, France.

(*.) Reprinted from "Hartl DM, Brasnu D. Les paralysies recurrentielles: connaissances actuelles et traitements. Annales d'Otolaryngologie et de Chirurgie Cervico-Faciale 2000; l17:60-84." [C] editions Masson.