A radiologic review of hoarse voice from anatomic and neurologic perspectives

The larynx, also known colloquially as the voice box, resides between the anatomic oropharynx and trachea and is subdivided into the supraglottis, glottis, and subglottis. It is comprised of six cartilaginous structures, three which are single—thyroid, cricoid, and epiglottis—and three which are paired—arytenoid, cornicuate, and cuneiform. The glottis contains the true vocal cords, which are double-layered membranous folds which attach to the arytenoid cartilages on either side. The vocal cords are responsible for vocal production via vibration as exhaled air passes through the glottis. The intrinsic musculature is responsible for vocal cord and epiglottic movement, while the extrinsic group provides support to the larynx. The primary intrinsic muscles responsible for vocal cord movement are the cricothyroid, posterior cricoarytenoid, lateral cricoarytenoid, transverse arytenoid, thyroarytenoid, and vocalis (Fig. 1). Aside from the cricothyroid, which is innervated by the superior laryngeal nerve, the muscles of the vocal cords are innervated by the recurrent laryngeal nerve; both nerves are branches of the vagus nerve.

The vagus nerve (CN X) originates from the nucleus ambiguous, located in the medulla oblongata between the medullary pyramid and the inferior cerebellar peduncle (Fig. 2). Upon leaving the medulla, the vagus nerve courses through the lateral cerebellomedullary cistern and exits the skull base via the pars vascularis of the jugular foramen. It then descends the neck within the carotid sheath, coursing between and posterior to the carotid artery (lateral) and internal jugular vein (medial), and enters the superior mediastinum near the aortic arch and origin of the great vessels.

The recurrent laryngeal nerve (RLN) branches anteriorly off the vagus nerve in the mediastinum. Due to the asymmetry of the aortic arch, the course on either side is slightly different; the right RLN loops posteromedially under the right subclavian artery, while the left RLN loops posteromedially under the arch itself, through the aortopulmonary window. Both then ascend along the tracheoesophageal groove toward the larynx (Fig. 3).

Vocal cord paralysis (VCP) occurs when there is impaired function of either CN X or the RLN. This can occur secondary to multiple disorders which can be characterized by etiology (Table 1) and location (Table 2). Most cases are unilateral and due to a compressive mass, although up to one-third may be bilateral [2]—usually seen in iatrogenic/traumatic injury, but also seen with neurogenic disorders such as amyotrophic lateral sclerosis or closed head injury which can have multifocal lesions. Up to 40% of people with unilateral VCP are asymptomatic at the time of diagnosis. VCP may be accompanied by co-symptomatology of other lower cranial neuropathies depending on the location of the lesion—a mass located from the brainstem to the jugular foramen may present with VCP along with dysphagia due to loss of pharyngeal sensation (CN IX dysfunction) and/or denervation of the trapezius and sternocleidomastoid muscles (CN XI dysfunction), while a mass within the carotid sheath may present with the aforementioned, as well as tongue atrophy and deviation (CN XII dysfunction).

The laryngeal ventricles of Morgagni are paired structures of the larynx residing between the vocal and vestibular folds; the saccule is an appendageal diverticulum extending vertically from the ventricle between the vestibular fold and the thyroid cartilage, and is responsible for producing mucus that lubricates the vocal cords. Cystic dilatation of the saccule is termed a laryngocele, which may be developmental due to failure of regression after birth, and can also be seen in persons who experience high throat pressures, such as wind musicians, glass blowers, and those with excessive coughing [3]. These are often asymptomatic and found incidentally, but occasionally can present with hoarseness and stridor. Symptoms may be episodic due to intermittent filling and resultant one-way valve effect, or due to infection [4].

On imaging, a laryngocele appears as a thin-walled air- or fluid-filled space lateral to and communicating with the laryngeal lumen; the wall will demonstrate absent or minimal enhancement [5] (Fig. 4). Types of laryngoceles are defined by their relationship to the thyrohyoid membrane—internal and external/mixed [4]. Internal (or simple) laryngoceles are confined to the paraglottic space of the supraglottis, and are found lateral to the thyrohyoid membrane. If the laryngocele extends superiorly through the thyrohyoid membrane into the submandibular space, then it is termed a mixed laryngocele. An external laryngocele does not have an internal component and is rare; it may present as a lateral neck mass and can be intimately associated with the superior laryngeal nerve.

Repetitive insult to the laryngeal mucosa—such as with singing, yelling, smoking, gastroesophageal reflux—can lead to a number of benign inflammatory lesions, especially of the vocal cords. Although related in etiology, these lesions differ in mechanics of origin and appearance [6, 7]. Vocal cord nodules occur bilaterally and symmetrically at the midpoint of the cords due to high shear force during apposition and subsequent remodeling, similar to a callous. A laryngeal polyp forms after trauma to the submucosal capillary bed, with subsequent exudative inflammatory changes and outpouching of the mucosa (Fig. 5). A laryngeal cyst can appear similar to a laryngocele and may be of two types: retention, due to glandular obstruction and resultant trapped secretions, or epidermoid, which are congenital or due to trauma. Reinke’s edema is not a discrete lesion, but is diffuse edema of the vocal cords due to fluid accumulation between the membranous folds (Reinke’s space). These lesions do not warrant imaging as they are readily seen by laryngoscopy; however, they may be seen incidentally on head and neck imaging obtained for other reasons [8].

Squamous cell carcinoma (SCC) is by far the most common (98%) primary tumor of the larynx, and can also secondarily involve the larynx when the primary is elsewhere in the oropharynx. SCC generally occurs in men over 50 and was previously associated with smoking and alcohol abuse; however, SCC related to human papilloma virus (HPV) is now more prevalent, and is seen in the younger population with better prognosis. Classification is based on subsite of location in relation to the glottis—supraglottic (20–30%), glottic (50–60%), subglottic (5%), and transglottic (spanning two or more subsites)—and presentation depends on the subsite involved. Glottic lesions often present earlier with dysphonia or aspiration, while subglottic lesions typically present with dyspnea and/or stridor. Purely supraglottic lesions are asymptomatic, thus usually present later with symptoms due to lymphadenopathy or trans-spatial spread, such as tender neck mass, sore throat, dysphagia/odynophagia, or referred ear pain.

As laryngoscopy can be performed by the otolaryngologist, imaging is used for staging rather than diagnosis. SCC appears as an enhancing soft tissue mass on both computed tomography (CT) and MRI; either may be used for disease surveillance. The lesion is usually isoattenuating to the surrounding mucosa on CT, but may be identified due to architectural distortion and loss of tissue planes (Figs. 6 and 7). On MRI, the lesion is usually T1-hyperintense, T2-hyperintense, and demonstrates high STIR signal due to edema. Positron emission tomography (PET) may be used in conjunction to identify metastatic lymphadenopathy or post-treatment occurrence.

Small lesions may be treated locally with laser or radiation ablation; larger tumors usually necessitate laryngectomy and radiation therapy, although supraglottic tumors without vocal cord fixation may be treated with voice-sparing surgery. Compared to similarly staged SCC elsewhere in the head and neck, laryngeal SCC portends a better prognosis, but 15–20% will develop a second primary site SCC that is most certainly fatal.

Respiratory papillomatosis is caused by human papilloma virus (HPV); it has a propensity for the larynx but can also occur elsewhere in the airway [9]. It is the most common laryngeal tumor in children, but can be seen in adults. In juvenile-onset respiratory papillomatosis, HPV is transmitted from a mother to her child during passage through the birth canal; an infected mother has a 1% risk for transmission to her child. In adult-onset respiratory papillomatosis, transmission is thought to occur via oral sex [10, 11]. The most common strains in respiratory papillomatosis are HPV 6 and HPV 11; however, HPV 16 carries the highest risk for malignant transformation. Respiratory papillomatosis results in airway obstruction, which may be due to either blockage by an intraluminal lesion or infiltration of the structural tissues; tissue infiltration can also result in dysphonia due to morphologic alteration of the laryngeal components. Imaging characteristics are nonspecific; intraluminal lesions tend to be sessile and verrucoid, while infiltrative lesions appear as irregular soft tissue thickening. Lesions tend not to enhance avidly or at all; however, they will demonstrate fluorodeoxyglucose (FDG)-avidity on PET (Fig. 8). Treatment is antiretroviral therapy and surgical resection, possibly complemented by airway reconstruction in cases of wide resection necessitated by large lesions or extensive infiltration; as implied by the name, however, papillomatosis tends to recur [12].

Amyloidosis is the extracellular deposition of a variety of fibrils that are composed of various low molecular weight subunits of proteins that are found normally in blood serum; this abnormal tissue is called amyloid. Laryngeal amyloidosis is a rare entity, accounting for only 1% of laryngeal masses [13, 14]. Amyloidosis is a slow progressive process that results in hoarseness due to tissue infiltration by amyloid, which alters the structure and mechanics of the vocal apparatus. Deposition in laryngeal tissue is most often a localized process, but systemic amyloidosis with laryngeal involvement is usually due to a monoclonal plasma cell dyscrasia. Imaging findings in amyloidosis are nonspecific and result from architectural distortion of laryngeal structures due to tissue infiltration (Fig. 9). Treatment consists of endoluminal microsurgical resection or debulking, and often requires multiple sessions due to extent and recurrence [15].

The dentato-rubro-olivary tract, also known as the myoclonic or Guillain-Mollaret triangle, connects the inferior olivary nucleus (ION) in the medulla to the ipsilateral red nucleus in the midbrain and contralateral dentate nucleus in the cerebellum via the central tegmental, olivocerebellar/olivodentate, and dentorubral tracts. A lesion disrupting any of these pathways can result in deafferentation of the inferior olivary nucleus, either contralateral (lesion of the superior or inferior cerebellar peduncle) or ipsilateral (lesion of the central tegmentum). Due to trans-synaptic degeneration, there is cytoplasmic vacuolar degeneration, along with glial hypertrophy and astrocytic proliferation, initially resulting in enlargement of the ION; this hypertrophy is unique to these lesions, although the olive eventually becomes atrophic. Often the offending lesion is hemorrhagic, but may be ischemic or demyelinating [16].

The classic clinical finding in hypertrophic olivary degeneration (HOD) is palatal myoclonus, which is involuntary rhythmic movement involving the oropharynx, typically the levator palatini but may also include the larynx; ocular myoclonus and tremor or ataxia of the upper extremity may also be present. Symptoms usually present several months after the primary insult, but the patient may also be asymptomatic.

MRI is characterized by distinct phases of progression. The ION initially appears normal, then develops T2-hyperintensity after 1 month. After several months to a few years, the ION remains T2-hyperintense and also becomes hypertrophic; this hypertrophy eventually resolves, and the ION becomes atrophic. Contrast enhancement is atypical (Fig. 10).

The posterior inferior cerebellar artery (PICA) territory includes the inferior occipital surface of the cerebellar hemisphere and inferior vermis, while perforators off the intracranial vertebral (V4 segment) and basilar arteries supply the brainstem. Occlusion of the distal V4 segment or PICA can result in infarction of the lateral medulla (Fig. 11); this is characterized by a constellation of symptoms known as lateral medullary or Wallenberg syndrome [17]. Due to involvement of the nucleus ambiguus, patients will have ipsilateral bulbar weakness affecting the muscles of the soft palate and pharynx, resulting in dysphonia, dysphagia, and dysarthria. Other symptoms include autonomic dysfunction, ipsilateral Horner’s sign (meiosis, ptosis, enophthalmos, anhydrosis), ipsilateral facial and contralateral truncal/appendicular sensory disturbance, and vestibulocerebellar symptoms such as ataxia and vertigo (patients will “fall toward the lesion”).

Paragangliomas, or glomus tumors, are slow-growing vascular neoplasms that arise from the neural crest, histologically equivalent to pheochromocytomas of the adrenal gland, and are intimately associated with neurovascular structures. Although extra-adrenal paragangliomas are most common in the head and neck, they are rare, accounting for approximately 0.5% of all head and neck tumors. Most are sporadic, but with new discoveries of genetic markers, it is now thought that up to 30% may be familial. Incidence slightly favors women, and diagnosis is typically during the third through fifth decades. Most are benign, and malignancy is determined by the presence of metastatic lesions as there is no histologic difference. The vast majority are not vasoactive; symptomatology is usually determined by location.

Paragangliomas are named for their location of origin; the four most common primary sites in the head and neck are at the tympanic membrane (glomus tympanicum), jugular fossa (glomus jugulare), within the carotid sheath (glomus vagale), and at the carotid bifurcation (glomus caroticum or carotid body tumor). These tumors have the propensity for trans-spatial spread; for example, a glomus jugulotympanicum tumor involves both the skull base and middle ear. Glomus jugulare tumors tend to present as pulsatile tinnitus and hearing loss, although lower cranial nerve symptoms such as hoarse voice can be present as well. Glomus vagale tumors occur along the vagus nerve and are accompanied by VCP approximately half of the time.

Paragangliomas can be imaged with a number of modalities, including CT, MRI, and nuclear medicine studies [18]. The intimate association of glomus tympanicum and jugulare with structures of the temporal bone and skull base lends well to the use of CT; aggressive tumors may result in bony destruction with a moth-eaten appearance. On MRI, these tumors are classically described as “lightbulb bright” on T2-weighted imaging due to extensive vascularity and “salt and pepper” due to flow voids. Avid enhancement is seen with both iodinated and gadolinium contrast. Due to their avidity for tumors of neuroendocrine origin, a number of radionuclides can be used to identify paragangliomas: ¹¹¹In-pentreotide and 68Ga-DOTATATE for somatostatin receptor positivity, ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) for metabolic activity, and ¹²³I-metaiodobenzylguanidine (¹²³I-MIBG) and ¹⁸F-fluorodopamine (¹⁸F-FDA) for catecholamine production.

Schwann cells are the principle glial cells of the peripheral nervous system, myelinating peripheral nerves, while perineural fibroblasts comprise the perineurium, which is the connective tissue surrounding nerve fascicles. From these nerve sheath elements arise schwannomas and neurofibromas; these benign tumors can affect any peripheral nerve, including the cranial nerves [20, 21]. Benign nerve sheath tumors of the head and neck are mostly schwannomas, although neurofibromas are seen in syndromic cases such as neurofibromatosis type I. These tumors can be difficult to distinguish from each other based on clinical or imaging examination alone—both will present as a well-circumscribed encapsulated mass with avid (usually homogenous) enhancement (Fig. 17)—so these entities are considered together when forming a differential diagnosis. They can be differentiated from paragangliomas by their intimate association with the peripheral nerve and lack of flow voids.

Of the cranial nerve schwannomas, the vast majority (90%) are of the vestibular branch of the vestibulocochlear nerve (CN VIII), but can also affect the trigeminal nerve (CN V), facial nerve (CN VII), or lower cranial nerves (CN IX-XII). Vagal nerve schwannomas have a slight female predominance and usually present in middle age, although they can occur at any age. These tumors most often present as an asymptomatic slow-growing palpable neck mass; however, the second most common presentation is hoarseness. The most specific sign for vagal nerve schwannoma is paroxysmal cough induced by palpation or manipulation of the mass, due to stimulation of the vagal nerve [22]. Treatment is surgical resection; preservation of the vagus nerve can be difficult; however, schwannomas tend to be more easily resected since they arise from the myelin surrounding the entire nerve and are often eccentric to the nerve itself, rather than neurofibromas which arise from the perineurium surrounding individual fascicles and thus can be entwined in the fibers [20, 22].