Paragangliomas of the head and neck are a unique collection of rare neoplasms. Their neuroendocrine origin necessitates a unique management approach, requiring comprehensive genetic and biochemical assessment in addition to a more standard tumor work-up. A variety of surgical and radiotherapeutic strategies can be employed to manage appropriately selected cases. However, many of these tumors exhibit an indolent natural history, progressing slowly over a patient’s lifetime. Consequently, a conservative observational strategy is often efficacious, limiting treatment-associated morbidity and helping optimize patient quality of life.
38 Paragangliomas of the Head and Neck
The human paraganglionic system comprises a highly specialized neurosecretory epithelium (composed of chromaffin cells) found within the adrenal medulla and along neuronal/vascular adventitia within extra-adrenal paraganglia.1 , 2 , 3 All elements of this system share a common neural crest origin and are characterized by catecholamine production and storage.4 A clear link exists to the autonomic nervous system, with sympathetic paraganglia generally found in the abdomen and thorax, whereas those of parasympathetic origin cluster exclusively within the neck and skull base.5 The most well-known head and neck sites are the carotid bifurcation (carotid body), jugular foramen, and tympanic plexus. However, paraganglionic tissue may also be found along the course of the vagus and glossopharyngeal nerves, including within the larynx, nasal cavity, orbits, and trachea.1 , 3 , 6
Extra-adrenal paragangliomas are rare neuroendocrine neoplasms, which comprise only 10% of all chromaffin cell tumors (with pheochromocytomas encompassing the other 90%); only about 3% of these lesions occur in the head and neck.2 , 7 Although catecholamine hypersecretion is a unique concern with paragangliomas, this is an uncommon concern within the head and neck (only 3–4% of cases) because of their parasympathetic origin.8 As a consequence, they most commonly present as benign, slow-growing, and hypervascular tumors with potential symptomatology linked to anatomical site of origin.7 , 8 Although surgical resection has been the historic treatment of choice for all paragangliomas, the association with key neurovascular structures is associated with significant morbidity (e.g., speech/swallowing disorder, severe blood loss, not insignificant stroke risk) and has led to an ever more complex treatment algorithm that increasingly emphasizes the potential role of observation.6 , 7
38.2 Incidence and Epidemiology
The rarity of head and neck paragangliomas (HNPGs) makes the true disease incidence rate somewhat difficult to ascertain. Most estimates suggest an incidence that ranges from 1 in 30,000 to 100,000 within the general population.8 , 9 , 10 , 11 , 12 They are generally most common in the fourth to sixth decades of life and show a strong female preponderance (roughly 2/3:1).3 , 7 , 13 Familial and sporadic variants exist. Multicentric lesions (occurring either synchronous or metachronous with the primary presenting tumor) occur in 10 to 20% of patients, but this phenomenon is far more common in those who have documented hereditary disorders.6 , 14 , 15 Because all HNPGs are derived from the same neural crest–derived cellular line, they share a unified pathologic appearance.3 , 14 , 16 Accordingly, they are classified on the basis of anatomical site of origin, with carotid body (CBP), jugulotympanic (JTP), and vagal (VP) paragangliomas the most common (Table 38.1).2 , 7 , 9 , 17
Several genetic mechanisms are known to play a key role in tumorigenesis and are discussed hereafter. Although some of these mechanisms may play a contributory role even in sporadic (nonfamilial) HNPGs, the underlying pathophysiologic basis for HNPGs remains poorly defined. There is however a key association, between chronic hypoxic states and tumor development (specifically at the carotid body).1 , 18 , 19 This was first noted in the 1960s and 1970s among Peruvians, who live at high altitudes in the Andes Mountains (2,105–4,350 meters above sea level).9 , 20 , 21 Reduced atmospheric oxygen seems to provoke carotid body hyperplasia; particularly large and heavy carotid bodies are thus common in this population. The prevalence of paragangliomas is correspondingly elevated, being roughly 10 times more frequent than among those living at sea level.9 , 19 Although the mechanism underlying chronic hypoxic stimulation is not clearly understood, this effect has been confirmed for other medical conditions, including cystic fibrosis, cyanotic heart disease, and central alveolar hypoventilation, in which the prevalence of HNPG is also increased.9
38.2.1 Genetics of Head and Neck Paragangliomas
Advances in molecular genetics over the past thirty years, have identified some genetic driver event as a cause of more than 50% of pheochromocytomas and paragangliomas (Table 38.2).22 Inherited paraganglioma syndromes, for example, are now known to be associated with germline mutations in one of at least 12 different genes. More recently, somatic mutations have been identified in related genetic elements within sporadic paraganglioma cases.10 , 13 , 22 , 23 How these nonheritable mutations are acquired, and their association with the pathogenesis of sporadic paragangliomas, is a particularly intriguing new area of study.4 , 23
Hereditary paragangliomas, can be broadly grouped into two clusters on the bases of gene expression profiles.22 , 24 The first cluster includes mutations in the von Hippel-Lindau (VHL), succinate dehydrogenase (SDH) subunits, and hypoxia-inducible factor 2A (HIF2A) genes.22 These genetic elements are involved in the hypoxia–angiogenesis pathway, acting to modulate a transcription factor known as HIF, which is typically upregulated within oxygen-sensitive tissue in response to hypoxia. Mutations in VHL, SDH, or HIF2A thus seemingly create a state of pseudohypoxemia, inducing a persistent angiogenesis (and thus tumorigenesis), a condition that can be presumed to be similar to that experienced by individuals living at high altitude.9 , 19 , 22 , 24 The second genetic cluster is more heterogeneous and is rarely associated with HNPGs. It includes genes associated with specific protein signaling [neurofibromin 1 (NF1), RET], mitogenesis [MYC-associated factor X (MAX)], and protein trafficking (TMEM127) pathways.4
Mutations mapped to the succinate dehydrogenase gene family (SDHA, SDHB, SDHC, SDHD, and SDHAF2) are the cause of a group of familial paraganglioma syndromes (known as PGL1–5), most specifically associated with HNPGs.7 , 10 , 25 , 26 , 27 These genes encode subunits of mitochondrial complex II, which is involved in the aerobic electron transport chain and Krebs cycle. The inheritance pattern of SDH mutations is somewhat unique. Although an autosomal dominant pattern is exhibited, there is both age-dependent penetrance and almost 100% maternal imprinting.10 , 28 , 29 Accordingly, phenotypic expression of the disease state is dependent on patient age, with the average presentation in the third decade of life.28 , 30 Maternal imprinting is an epigenetic phenomenon in which alleles inherited from the mother become inactivated. Consequently, an affected father has a roughly 50% chance of disease transmission, whereas affected mothers can only transmit inactivated genes, which could be reactivated in the next generation.10 , 28 , 29
Contrary to the previously dogmatic 10% rule, current evidence suggests that roughly 30% of all paragangliomas possess a familial germline mutation.4 , 23 , 24 , 28 , 29 , 30 The PGLs in particular are generally characterized by five key features: (1) a positive family history, (2) young age (≤ 45), (3) preceding or simultaneous pheochromocytoma, (4) multiple paragangliomas, and (5) male gender.13 , 30 It is interesting to note this male predominance, which is contrary to the female predominance observed with sporadic tumors. The precise disease phenotype is, however, dependent on the exact succinate dehydrogenase defect present. SDHD mutations (the leading cause of familial HNPGs), for example, are associated with multifocality. SDHB mutations, by contrast, are associated with a particularly high rate of malignant paraganglioma (> 30% risk).2 , 7 , 13 , 24
Because the price of genomic screening costs from $1,000 to $2,000, a targeted genetic screening approach has generally been advocated based on the five key clinical predictors outlined in the preceding paragraph.25 , 30 More recent advances in molecular testing have led others to advocate for universal genetic testing (targeting SDHB, SDHC, SDHD, and VHL) given a > 10% likelihood of heritable mutations in otherwise presumed sporadic cases.23 , 25 , 26 , 30 Another important consideration is the rapidly growing list of associated somatic and germline mutations now linked to these tumors, with potential clinically relevant consequences for patient and family members alike. Genetic counseling should at the very least be advocated in the majority of cases to help guide this process.25
38.2.2 Malignant Paraganglioma
Malignant paragangliomas are a rare entity responsible for 3 to 5% of these lesions within the head and neck.13 Interestingly, the likelihood of malignancy is dependent on tumor anatomical localization, being particularly likely for vagal (16–19%) lesions.6 , 31 Diagnosis is generally made on the basis of metastatic disease considering the difficulty distinguishing from benign lesions on histopathology.6 , 13 , 31 , 32 However, the presence of significant invasion of local structures is considered to represent malignant disease by certain authors.33 Although isolated regional cervical lymph node spread is most common (55–70% of cases), distant disease dissemination (when occurring) is most common to bone, lung, and liver.13 , 32 Interestingly regional spread is particularly common for malignant carotid body tumors (94% of cases), whereas distant metastasis is more common for malignancies at other sites.31 , 32 Overall survival for patients receiving treatment (surgery ± adjuvant radiotherapy [XRT]) for disease isolated to the head and neck is high—roughly 80 to 90% at 5 years.31 , 32 Treatment results become disappointing, however, in cases involving distant metastatic spread.13 , 31 , 32
Paragangliomas and pheochromocytomas are highly vascular neuroendocrine tumors that share a uniform histopathologic appearance.1 , 3 , 14 They are generally surrounded by a pseudocapsule that may show evidence of capsular penetration/vascular invasion, but these findings are not diagnostic of malignancy.14 Architecturally, a classic “Zellballen” pattern is described, with tumor cells arranged in round oval nests that vary in size (Fig. 38.1). Two cell types are present the predominant chief cells (type I epithelioid cells) and sustentacular cells (type II supporting cells).1 , 3 , 14
The chief cells are of neuroendocrine lineage and thus possess catecholamine-containing granules. They are often characterized by nuclear enlargement and a varying cytoplasmic component that ranges from granular eosinophilic to deeply basophilic.14 These cells stain positive for neuroendocrine histochemical markers such as chromogranin, synaptophysin, NSE, and CD56. The sustentacular cells are a group of stromal cells that act like neural glia in creating a supportive/vascular network around chief cell nests.14 These cells have a unique immunohistochemical pattern and are uniformly s100 protein–positive.
Catecholamine synthesis is an important hallmark of the paraganglionic system. This biosynthetic pathway uses a series of enzymatic catalysts to convert precursor substrates (phenylalanine and tyrosine) into functional catecholamines (dopamine, norepinephrine, and epinephrine).4 Catecholamine catabolism also uniquely occurs within paraganglionic tissue (including tumors) through the action of catecholamine-O-methyl transferase, producing a group of metabolites known as metanephrines.4 , 34 HNPGs rarely cause symptomatic catecholamine secretion (< 4%), because the parasympathetic derived tissue lacks the downstream enzymatic mechanism needed to produce epinephrine.4 , 6 , 13 , 14 , 35 These tumors do, however, tend to produce dopamine, so the associated metanephrine methoxytyramine may be detectable even in clinically silent tumors.34 Despite the rarity of hypersecretory HNPGs, the potential life-threatening nature of functional tumors and risk of multicentricity lead most authors to recommend biochemical screening by measurement of 24-hour urinary or plasma metanephrines.4 , 33 , 34
Because most HNPGs are benign, traditional staging systems are not applicable. Site-specific disease classifications have, however, been devised, particularly as a means of guiding surgical resection (Table 38.3). Since 1971, carotid body tumors have been stratified based on the Shamblin system.6 , 7 , 13 Shamblin class I tumors simply splay (but do not invade the carotid vessels), whereas Class III tumors completely encase both the internal and external carotid artery (Fig. 38.2). Higher-class lesions thus require more extensive resection (and potentially carotid artery reconstruction), leading to an increasingly higher risk of treatment-related morbidity. For JTPs, two separate staging systems are commonly referenced: Fisch-Mattox and Glascock-Jackson.13 , 36 The Fisch-Mattox system incorporates all these lesions into a single continuum, whereas Glasscock-Jackson denotes tympanic and jugular bulb tumors separately. No specific staging system is in use for VPs.
The obvious challenge posed by HNPGs is that of balancing the potential treatment associated with tumor-specific morbidity in a population of patients who most often present with benign asymptomatic disease. The concern is, of course, unrestrained growth and local invasiveness, which may lead to both progressive cranial nerve and neurovascular deficits.6 , 7 Similar morbidities are, however, caused by curative treatment protocols, with a > 10% likelihood of potentially serious adverse outcomes reported in most series.6 , 7 , 37 Unfortunately, a uniformly accepted treatment algorithm does not exist. A detailed pretreatment clinical work-up, knowledge of patient-specific comorbidities/risk tolerance, and an understanding of disease natural history are thus essential to guide management decision making.
During the initial clinical encounter, disease symptomatology particularly related to the function of potentially at-risk cranial nerves (VII–XII), must be assessed. In most situations, anatomical (and potentially functional imaging) can be used for accurate tumor categorization, with biopsy only rarely indicated.6 , 38 , 39 , 40 Contrast-enhanced MRI is particularly useful, as these tumors exhibit an almost pathognomonic “salt-and-pepper” appearance on T2-weighted imaging, characterized by a high degree of contrast enhancement and vascular flow voids (Fig. 38.3).6 , 40 In cases being considered for surgical management, the addition of MR/CT angiography is advocated to better map the association between the tumor and surrounding vascular anatomy. The use of functional imaging techniques (e.g., 123I/131I-metaiodobenzylguanidine scintigraphy,111 In-DTPA pentetreotide scans, SPECT/CT) are also proposed, both as a means of defining the individual tumor natural history and as a screen for metachronous paragangliomas.39 , 40 However, the optimal role of these various techniques remains under investigation.24 , 39 , 40
The concept of applying a noninterventional approach to HNPGLs is based on a presumed indolent natural history. This was first outlined by van der May et al from Leiden University, in the Netherlands. They reviewed 52 cases of JTPs with prolonged follow-up (mean 13.5-year follow up); although most were treated surgically (radical vs. subtotal resection), 13 cases were observed.41 None of the patients died from tumor or developed distant metastasis, but the likelihood of cranial nerve deficits was two times higher in the surgical cohort.41 , 42 A more recent update by this group reviewed 48 HNPGs (in 26 patients, as a result of familial disease variants), managed initially with observation. Further reinforcing the indolent nature of these tumors (at least in the Dutch population), they calculated a slow tumor growth rate (0.83 mm/y) with a median time to tumor doubling of 10.2 years.43 However, the growth rate was incredibly variable, with doubling time ranging from 0.6 to 21.5 years.43 Rapid growth was most common in intermediate-sized tumors (0.8 to 4.5 mL), indicating size as a potential prognostic factor of disease progression. They propose a volume increase of ≥ 20% (a criterion reached by 60% of tumors) to be a clear sign of tumor enlargement and potential threshold for intervention.43
A collection of studies has further elaborated on the potential utility of observation for jugular paragangliomas in particular.42 , 44 A combined 39 patients (Fisch type C and D) have been managed using this approach, with only 3 eventually requiring treatment. During the period of follow-up, tumor growth occurred in only 30 to 40% of cases, with a slow tumor growth rate (< 1 mm/y) reported in most patients. However, this cohort, by and large, included older patients (> 65) with generally < 10 years of follow-up, limiting the potential applicability to younger individuals.42 , 44
The Vanderbilt university group has also reported a series of 47 cervical paragangliomas (in 43 patients), managed without intervention. Similar to the other series, an indolent natural history was noted, with tumor growth in only 38% of cases and a mean growth rate of 0.2 cm/year.45 None of these cases required aggressive management, and no new cranial nerve deficits arose during the observation period—which, again, was relatively short-term (mean 5 y).45