Paragangliomas (PGLs) are tumours originating from neural crest-derived cells situated in the region of the autonomic nervous system ganglia. Head-and-neck PGLs (HNPGLs) originate from the sympathetic and parasympathetic paraganglia, most frequently from the carotid bodies and jugular, tympanic and vagal paraganglia, and are usually non-catecholamine secreting. Familial PGLs are considered to be rare, but recently genetic syndromes including multiple PGLs and/or phaeochromocytomas have been more thoroughly characterised. Nowadays, genetic screening for the genes frequently implicated in both familial and sporadic cases is routinely being recommended. HNPGLs are mostly benign, generally slow-growing tumours. Continuous growth leads to the involvement of adjacent neurovascular structures with increased morbidity rates and treatment-related complications. Optimal management mostly depends on tumour location, local involvement of neurovascular structures, estimated malignancy risk, patient age and general health. Surgery is the only treatment option offering the chance of cure but with significant morbidity rates, so a more conservative approach is usually considered, especially in the more difficult cases. Radiotherapy (fractionated or stereotactic radiosurgery) leads to tumour growth arrest and symptomatic improvement in the short term in many cases, but the long-term consequences are unclear. Early detection is essential in order to increase the chance of cure with a lower morbidity rate. The constant improvement in diagnostic imaging, surgical and radiation techniques has led to a safer management of these tumours, but there are still many therapeutic challenges, and no treatment algorithm has been agreed upon until now. The management of HNPGLs requires a multidisciplinary effort addressing the genetic, surgical, radiotherapeutic, oncological, neurological and endocrinological implications. Further progress in the understanding of their pathogenesis will lead to more effective screening and earlier diagnosis, both critical to successful treatment.
Paragangliomas (PGLs) are rare tumours originating from paraganglia – small groups of neuroendocrine cells arising from the autonomic nervous system ganglia. The sympathetic paraganglia are mostly located along the sympathetic nerve chains bordering the vertebrae and in the pelvis, while the parasympathetic ones are primarily located in the head and neck and less frequently located in the thorax or pelvis. Tumours arising from the parasympathetic paraganglia are usually non-chromaffin and only rarely secrete catecholamines, when compared with their sympathetic counterparts (Barnes et al. 2004).
Head-and-neck PGLs (HNPGLs) are rare tumours, representing 0.012% of a large oncological surgical series (Lack et al. 1977); the estimated clinical incidence is 1/100 000 patients per year (Baysal 2002). The classical main sites of origin are as follows: carotid bodies (at the bifurcation of the common carotid artery); jugular paraganglia (close to the jugular bulb) and tympanic paraganglia (in the middle ear) – usually considered together (JTPGLs); and vagal paraganglia (along the vagus nerve). The most frequently found are the carotid body tumours (CBTs) and the least frequent are those arising from the vagus paraganglia (VPGLs). The relative frequencies across series in the literature vary widely: in a large reported series of 204 HNPGLs, 57% were CBTs, 30% JTPGLs and 13% VPGLs (Erickson et al. 2001).
Increasingly, due to their association with phaeochromocytomas and new genetic data, these tumours are being diagnosed by endocrinologists and oncologists, but much of the published literature is in more specialised ENT or surgical journals. We think that it will be useful to survey current data on the management of these tumours in order to assist clinicians in advising their patients on the most appropriate therapy.
HNPGLs were occasionally described in rare conditions such as von Hippel–Lindau disease (Zanelli & van der Walt 1996, Gaal et al. 2009), multiple endocrine neoplasia type 2 (Boedeker et al. 2009a), neurofibromatosis (DeAngelis et al. 1987) and Carney's triad (Carney et al. 1977). Until quite recently, most PGLs were considered sporadic, unless there was a positive family history (FH) or if co-morbidities characteristic of the known genetic syndromes were present. However, many newly characterised genes have been found to be mutated in many apparently sporadic PGLs. Currently, four genetic PGL syndromes are being described, all with autosomal dominant transmission. Three of these syndromes are associated with germline mutations in the gene complex encoding succinate dehydrogenase (SDH): PGL-1, PGL-3 and PGL-4, caused by mutations in SDHD, SDHC and SDHB respectively. All these mutations predispose to PGLs in all locations and/or phaeochromocytomas. HNPGLs are frequent in the PGL-1 syndrome due to SDHD mutations and are frequently multifocal and also in the PGL-3 syndrome due to SDHC mutations; they are more rare but more frequently malignant in the PGL-4 syndrome secondary to SDHB mutations. In the PGL-1 syndrome, genomic imprinting has consistently been described; the maternally derived allele is imprinted, so this means that only mutations inherited from the father are pathogenic (Niemann et al. 1999, Baysal et al. 2000, Astuti et al. 2001, Gimenez-Roqueplo et al. 2012). The causal mutation of the PGL-2 syndrome has only recently been described (SDHAF2 gene; Hao et al. 2009), is very rare (Bayley et al. 2010) and its transmission is also consistent with genomic imprinting (Kunst et al. 2011).
Apparently sporadic PGLs harbour SDH-related mutations in a significant proportion of the cases (30.6%); the likelihood of having a SDH mutation is best predicted by a positive FH, multicentricity and a previous phaeochromocytoma and, to a lesser extent, by young age, male gender and malignancy (Neumann et al. 2009). Therefore, sequential genetic screening guided by clinical predictive factors has been recommended, although parallel sequencing of multiple genes is carried out in some centres, including our centre (Young 2006, Burnichon et al. 2009). The loss of the normal immunohistochemical staining of paraganglial cells with antibodies against the SDHB protein has recently been found to be highly associated with germline mutations of all the SDH complex genes (van Nederveen et al. 2009). Greater experience is needed with this elegant potential screening tool, but it promises to become a routine part of the screening as immunonegativity for the SDHB protein is consistently associated with SDH-related tumours and is not found in sporadic or non-SDH mutation-related cases. In an international study, the penetrance of gene mutations has been found to be 48% for SDHD at 30 years of age and 73% by 40 years of age, vs 29 and 45% respectively for SDHB (Benn et al. 2006). Thus, the high penetrance of SDHD mutations suggests that they are rarely found in patients presenting tumours beyond 45 years of age (Cascon et al. 2009). In a large UK study, the majority of the mutations missed using an age cut-off were in the SDHB gene; these authors suggested that if genetic testing for SDHB is limited to younger patients on cost grounds, it should be complemented with SDHB immunostaining (Jafri et al. 2013).
It has been suggested that genetic testing for VHL, RET and NF1 mutations may be considered if there is clinical suspicion (Boedeker et al. 2009a,b) and for SDHAF2 in high-risk patients negative for other SDH mutations (Bayley et al. 2010), while two other newly described genes (TMEM127 and MAX) are rarely implicated in HNPGLs (Gimenez-Roqueplo et al. 2012). However, as has been noted above, we routinely screen for all the genes currently implicated in HNPGLs other than NF1; the latter has a very large number of exons, and the clinical syndrome is (almost always) obvious.
HNPGLs rarely release catecholamines to produce a hypersecretory syndrome (<10%). In the vast majority of cases, they are discovered due to the mass effects dominated by the involvement of lower cranial nerves (CNs) IX and X; 10% are diagnosed incidentally (Erickson et al. 2001).
HNPGLs in the lower part of the neck (CBTs and some VPGLs) usually present as painless, sometimes pulsatile, neck masses. With further growth, they involve the lower CNs, leading to speech and swallowing deficits (hoarseness and dysphagia) and sometimes to aspiration (Miller et al. 2000, Offergeld et al. 2012). A preoperative CN deficit is frequently observed in VPGLs (25–36%) and JTPGLs (39–40%) and less so in CBTs (4–22%) (Powell et al. 1992, Netterville et al. 1998, Sajid et al. 2007, Neskey et al. 2011).
Intracranial extension is rare in CBTs (Rao et al. 1999) and more frequent in VPGLs – 22% (Netterville et al. 1998). JTPGLs are intracranial tumours, and it is often difficult to delineate their site of origin: tympanic tumours may extend towards the jugular bulb and posterior fossa, while jugular tumours can involve the temporal bone and extend into the middle ear. Intracranial invasion and involvement of the CN adjacent to the jugular foramen can lead to pulsatile tinnitus, an ear mass, hearing loss, pain and vertigo as major presenting symptoms in JTPGLs and high VPGLs (Cummings et al. 1984, Persky et al. 2002, Offergeld et al. 2012).
HNPGLs are most frequently diagnosed in middle-aged adults (mean age 41–47 years; Erickson et al. 2001, Jackson et al. 2001, Papaspyrou et al. 2009). Genetic cases are more than a decade younger (Burnichon et al. 2009). A FH should be sought in all the cases; positivity is highly variable among series, but can reach over 80% in populations with a high frequency of founder SDHD mutations (Hensen et al. 2011). A positive FH increases the risk of multifocality: up to 78% of the cases with a positive FH and 17–37% of the unselected cases are multicentric (Jackson et al. 1990, Netterville et al. 1998, Erickson et al. 2001, Plukker et al. 2001, Sajid et al. 2007, Papaspyrou et al. 2009).
Malignancy can only be ascertained in the presence of distant metastases. None of the markers in a large set of putative markers of malignancy can reliably predict malignancy, and tumour size is still the major indicator of risk, as it is for phaeochromocytomas (Korevaar & Grossman 2011). The risk of malignancy is higher in SDHB mutation carriers, while multifocality is more frequent in SDHD germline mutations (Burnichon et al. 2009). In large series, evidence of malignancy in 3–5% of the HNPGL cases (Manolidis et al. 1999, Jafri et al. 2013) with a lower risk for CBTs and JTPGLs (2–6%) and a higher risk (16%) for VPGLs has been found (Kahn 1976, Kloppel 2003). Distant metastases may occur even after 16 years of follow-up (Lees et al. 1981), so long surveillance is critical for a precise estimate of the real incidence. The most common site of distant spread is the cervical lymph nodes (68.6%; Lee et al. 2002); other sites are the bone, lung and liver (Moskovic et al. 2010).
The overall 5-year survival in malignant cases is generally good (59.5–84%; Manolidis et al. 1999, Lee et al. 2002, Moskovic et al. 2010), but it becomes disappointing (11.8%) if cases with only local spread are excluded (Lee et al. 2002).
Shamblin classification of CBTs
The CBT classification of Shamblin et al. (1971) is still in use and shows a good correlation with surgical complications and outcome (Plukker et al. 2001, Luna-Ortiz et al. 2005, Makeieff et al. 2008). Tumours classified as class I have no or minimal attachment with the carotid arteries. Class II tumours surround the carotid arteries, partially encasing them. Shamblin class III tumours surround the vessels, adhering firmly over their whole circumference, so vessel resection and reconstruction are needed to attempt total tumour resection.
Fisch classification of JTPGLs
JTPGLs have been similarly classified by Fisch & Mattox (1988): class A JTPGLs are located along the tympanic plexus on the promontory, and class B tumours invade the hypotympanum, but do not erode the jugular bulb, as opposed to class C tumours (C1 destruction of the jugular bulb/foramen; C2 invasion of the vertical carotid canal; C3 invasion of the horizontal carotid canal and C4 invasion of the cavernous sinus). In class D tumours, besides the various degrees of invasion described for class C, intracranial extradural or intradural extension occurs (De1 and De2 intracranial and extradural invasion of up to 2 cm or more than 2 cm respectively; Di1, Di2 and Di3 intracranial and intradural extension of up to 2 cm, between 2 and 4 cm or more than 4 cm respectively).
Plasma or 24-h urinary metanephrine or catecholamine concentrations should be measured in all HNPGLs, as they can (rarely) be secretory (Erickson et al. 2001). If catecholamine excess is demonstrated, an extensive workup should be performed to assess the possibility of synchronous phaeochromocytoma/sympathetic PGLs. Exclusive dopamine secretion was considered to be very rare, usually silent and possibly associated with a trend towards increased aggressiveness or malignant potential (Eisenhofer et al. 2005), but recently higher rates (16.75–23%) of solely dopamine secretion or secretion of its metabolites have been described with these tumours (van Duinen et al. 2010, Van Der Horst-Schrivers et al. 2010). Chromogranin A is only rarely secreted (van Duinen et al. 2011). Thus, such markers are only useful in the follow-up of selected tumours.
Imaging is of paramount importance in patients with a clinical suspicion of HNPGLs or in individuals from affected families. Standard anatomical imaging (computerised tomography (CT) and magnetic resonance imaging (MRI)) is widely used as the initial evaluation method (see Figs 1, 2 and 3). CT has a lower sensitivity (Erickson et al. 2001), but accurately defines possible bone invasion. In MRI studies, PGLs show a characteristic ‘salt-and-pepper’ pattern (due to their vascularisation) and intense post-contrast enhancement. Doppler-flow ultrasonography can be useful for the diagnosis and follow-up of CBTs (Rao et al. 1999).
Angiography (digital-subtraction angiography (DSA) or MRI angiography) is a key method for diagnosis, defining the vascular anatomy preoperatively (tumour blood supply and extent of the involvement of the carotid arteries), ensuring safe preparation in the event of major vascular reconstructive surgery, demonstrating the patency of cerebral blood flow and identifying multicentric disease. MRI angiography is technically preferable and safer, but it has a lower sensitivity than DSA (van den Berg et al. 2000).
Whole-body screening with functional imaging studies can detect possible synchronous lesions in confirmed mutation carriers or those at a high-risk of familial disease (positive FH and young age). 123I-MIBG (metaiodobenzylguanidine) scintigraphy, despite its high specificity, has a low sensitivity for the detection of HNPGLs. Its use for standard evaluation of HNPGLs is limited, and it is more frequently employed to assess tumour avidity for the tracer if radionuclide therapy is planned. Somatostatin receptor (SSR) scintigraphy has a better sensitivity, lower than that of standard anatomical imaging but offers whole-body scanning (Timmers et al. 2012, Gimenez-Roqueplo et al. 2013).
Positron-emission tomography (PET) has been used intensively as it can be used for the examination of the whole body with improved resolution, and it can detect small and metastatic lesions. 18F-fluorodihydroxyphenylalanine (18F-DOPA) PET has a very good sensitivity for the detection of both unselected (Gabriel et al. 2013) and SDH-related HNPGLs (King et al. 2011). In a small group of HNPGLs of unknown genetic status, 68Ga-DOTANOC (DOTA-naphthyl-alanine conjugated with octreotide) PET/CT used for baseline evaluation is also significantly superior to CT, MRI and 131I-MIBG scintigraphy for the detection of multicentric disease or distant spread (Sharma et al. 2013).
Where available, 18F-DOPA–PET should be used as the first-line functional imaging method for the detection of suspected HNPGLs (while in sympathetic PGLs, 18F-fluorodopamine PET/CT is advocated as the first-line detection method). If unavailable, 18F-FDG- or 68Ga-DOTA-peptide-PET or SSR scintigraphy (if PET is not available) should be used to complement anatomical imaging studies (Timmers et al. 2009, Blanchet et al. 2011; see Figs 4 and 5).
The natural history of HNPGLs is estimated based on the surveillance of selected patients without significant symptoms and/or poor candidates for treatment. In a study, of the 47 presumed HNPGLs observed for 5 years, on average, 42% were stable and 38% slowly increased in size (mean annual growth 0.2 cm), while 20% decreased in size (any change in greatest dimension) (Langerman et al. 2012). Others have described a median growth rate of about 1 mm/year (0.3–5 mm) with a widely variable tumour doubling time (0.6–21.5 years), but overall 60% of the tumours increased by at least 20% of their original size over 1–8 years (average 4.2) (Jansen et al. 2000).
These results are generally considered reassuring, but the selection bias, short follow-up duration, lack of tumour-associated morbidity data and possible misdiagnosis or unrecognised malignant potential should be borne in mind.
The main treatment modalities for PGLs include surgery and radiotherapy (external-beam radiotherapy (EBRT) or stereotactic radiosurgery (SRS)). The respective roles of chemotherapy and peptide receptor radionuclide therapy (PRRT) are yet to be clearly defined. Owing to the relatively mild natural history of HNPGLs, a long follow-up duration is needed before reaching a correct conclusion about the efficacy of any treatment.
Whenever possible, complete surgical excision of the tumour is considered by many to be the favoured option of treatment in order to prevent morbidity associated with further tumour growth or later spread from an unrecognised malignant tumour. Overall, gross total resection (GTR) is achievable in 90–97% of the cases with a low surgical mortality rate (0–2.7%). However, surgery is associated with haemorrhagic, cerebrovascular and neurological risks (paresis of lower CN leading to swallowing and speech problems, aspiration, feeding tube or tracheostomy dependence, facial palsy and hearing loss; Erickson et al. 2001, Jackson et al. 2001, Kollert et al. 2006, Paris et al. 2006, Sevilla Garcia et al. 2007, Papaspyrou et al. 2009, Neskey et al. 2011, Obholzer et al. 2011). Adjuvant procedures to improve swallowing or hoarseness (e.g. vocal cord medialisation) are frequently used postoperatively, but complete rehabilitation of the complex neurological deficit is slow in younger patients and often impossible in the elderly (Netterville & Civantos 1993), and thus indications for surgery should always be considered with great care.
Particular surgical challenges are associated with different HNPGL locations
Carotid body tumours
Complete surgical resection is possible in the vast majority (85–100%) of CBTs and recurrences are rare (Lees et al. 1981, Gaylis et al. 1987, Rodriguez-Cuevas et al. 1998, Makeieff et al. 2008, Ma et al. 2009). However, surgical morbidity is still significant due to the high vascularity and proximity to essential neurovascular structures. Significant intraoperative bleeding occurs, especially in advanced tumours (mean blood loss of 2200 ml in Shamblin class III tumours; Plukker et al. 2001). To reduce bleeding, preoperative tumour embolisation has been advocated (Persky et al. 2002, Ulrich et al. 2009), but others have reported a lack of efficacy (Zeitler et al. 2010) and/or an increased risk of stroke (Westerband et al. 1998, Fruhmann et al. 2013), so its utility remains controversial. It is likely to be highly operator dependent and could be considered for large tumours, but only if there is appropriate local expertise.
The risk of transient ischaemic attack or stroke is higher in CBT surgery than in the surgery of tumours present in other locations. In an analysis of the postoperative outcome of 1181 patients, the cumulative incidence was found to be 6.3% (Anand et al. 1995), but was lower in more recent series, between 0 and 4.8% (Plukker et al. 2001, Luna-Ortiz et al. 2005, Sajid et al. 2007, Makeieff et al. 2008, Ma et al. 2009). A vascular surgeon should be an essential part of the surgical team as internal carotid artery injury occurs frequently (10–23%) and vessel reconstruction leads to significantly lower stroke and mortality rates vs ligation (Anand et al. 1995, Plukker et al. 2001). Advances in surgical techniques and a multidisciplinary approach have led to a dramatic overall decrease in surgical mortality rates – 1% in a recent UK multicentre study (Sajid et al. 2007) vs 3.2% in a literature review carried out in 1995 (Anand et al. 1995) and 12.82% in older series (Lack et al. 1977).
Early postoperative CN deficits (lower CN palsies or, rarely, Horner's syndrome) are frequent (19–50%), but a permanent deficit is less common due to progressive slow rehabilitation, 1–18% (Plukker et al. 2001, Patetsios et al. 2002, Persky et al. 2002, Sajid et al. 2007, Makeieff et al. 2008). In advanced cases (Shamblin classes II and III), the rate of permanent neurological deficit can be up to 38% (Luna-Ortiz et al. 2005). The Shamblin classification is significantly correlated with postoperative complications (Makeieff et al. 2008), intraoperative blood loss (Plukker et al. 2001) and vascular reconstruction need (Smith et al. 2006), so early detection is essential for safe management.
A particular complication in CBT surgery is blood pressure instability (as the carotid body physiologically functions as a baroreceptor): both hypotension (Kohler et al. 2004) and, with bilateral resection, severe resistant hypertension (baroreceptor failure syndrome) can occur (De Toma et al. 2000). Chemoreflex dysfunction (absence of a normocapnic hypoxic ventilatory response) is almost universal, but baroreflex dysfunction occurs inconstantly (Timmers et al. 2003).
The probability of surgical cure in VPGLs is also very high: GTR is possible in 92.3–100% of the cases with a low mortality rate (0–2.7%) (Urquhart et al. 1994, Netterville et al. 1998, Jackson et al. 2001, Kollert et al. 2006). In a large literature review on VPGLs and JTPGLs published in 2012, Suarez et al. (2013a) described an average mortality rate of 1.3% with a GTR rate of 93.3%.
However, the neurological risks are higher than those for CBTs: CN palsies are more common after surgery for JTPGLs and high-located VPGLs due to the tumoural involvement of the jugular foramen. An immediate postoperative CN palsy rate can be as high as 96 and 100% respectively for these tumours (Neskey et al. 2011). The most affected component is the vagus nerve itself: in most series, a postoperative vagal deficit is almost universal (92–100%), by either paresis or necessary sacrifice during surgery (Urquhart et al. 1994, Netterville et al. 1998, Jackson et al. 2001, Bradshaw & Jansen 2005, Kollert et al. 2006, Zanoletti & Mazzoni 2006).
Other new CN deficits postoperatively occur in 23–61% of the cases for nerves IX, XI and XII (mostly IX) and in 15–17% for the facial nerves (Urquhart et al. 1994, Jackson et al. 2001, Bradshaw & Jansen 2005, Kollert et al. 2006, Zanoletti & Mazzoni 2006, Lozano et al. 2008). Severe aspiration as a consequence occurs in 10.2% of the cases (Suarez et al. 2013a). The majority of patients need complex rehabilitation management regarding speech, swallowing and facial nerve deficits, but during follow-up, these deficits often recover partially (Netterville et al. 1998).
Other significant complications include cerebrospinal fluid (CSF) leak, stroke and meningitis, present in 2.6, 2.2 and 0.4% of the cases respectively (Suarez et al. 2013a).
Other authors have suggested that the observation of VPGLs is associated with a better outcome: new CN palsy occurs in 7.5% of the cases (vs 60% postoperatively in the same series) and a 5% increase in size can be observed over 8.5 years (Bradshaw & Jansen 2005). However, in this series, 75% were familial cases and over half asymptomatic (therefore possibly less advanced), although 2.5% developed metastases during follow-up.
Surgery is most challenging for JTPGLs as extensive exploration of the posterolateral skull base is required. In most studies, GTR has been achieved still in the majority of the cases (59–96%) with 0–5% mortality rate (no more than 2% in the most recent series), even for complex tumours. The GTR rate is generally lower in the Fisch C and D classes, and it decreases to 41% in class D tumours or 35% in those with a large intradural extension (Glassock et al. 1979, Jackson et al. 1990, Green et al. 1994, Gjuric et al. 1996, Briner et al. 1999, Moe et al. 1999, Forest et al. 2001, Tran Ba et al. 2001, Al-Mefty & Teixeira 2002, Saringer et al. 2002, Pareschi et al. 2003, Suarez et al. 2007, Huy et al. 2009).
A meta-analysis published in 2011 has computed a pooled estimate of tumour control with GTR and subtotal resection (STR) of 86 and 69% respectively (Ivan et al. 2011). In another review of the treatment results for JPGLs, the overall long-term tumour control irrespective of the degree of resection has been reported to be 78.2% with a 1.6% treatment-related mortality rate. The risk of recurrence after apparent GTR is 6.9% (Suarez et al. 2013a). However, in most surgical series, the follow-up duration is generally too short to allow all recurrences to be reported; with a longer follow-up duration (112 months), 18.8% can recur (Papaspyrou et al. 2009). The validity of pooling inhomogeneous data from different centres is questionable, but since large series are extremely rare, this approach offers some idea as to what to expect from surgery in these tumours.
The functional outcome following surgery is generally poor. At particular risk are the facial nerves (frequently mobilised in order to improve tumour exposure and removal) and hearing function. Immediate postoperative facial weakness is frequent, and long-term dysfunction is present in 14–33% of the cases (Briner et al. 1999, Huy et al. 2009). Hearing function either remains stable or deteriorates; it is subjectively improved in only 6–16% of the cases, while 39% may experience a deterioration (Glassock et al. 1979, Gjuric et al. 1996, Briner et al. 1999, Kunzel et al. 2012). Overall, up to 45.5% of the cases have some degree of hearing loss after surgery (Suarez et al. 2013a). The global risk of other postoperative CN deficits after GTR is also high: 38, 26, 40 and 18% of the cases for nerves IX, X, XI and XII respectively (Ivan et al. 2011). Adjuvant procedures (e.g. tracheostomy, feeding tube, vocal cord medialisation procedure and gastrostomy) are used frequently, with at least one procedure being needed in up to 30% of the cases (Gottfried et al. 2004). However, because CN deficits are common preoperatively in JTPGLs and trigger obscure compensation mechanisms, long-term feeding tube dependence is equally frequent after CBT and JPGL surgery (Neskey et al. 2011). Overall, the risk of a new permanent deficit is still low in complex, very advanced JTPGLs, and a high GTR rate (85%) can be achieved even in such cases (Al-Mefty & Teixeira 2002). Normal activity after JTPGL surgery resumes slowly, being observed in 72% of the patients at 6 months and 97% after 1–2 years, reflecting the slow but progressive return of neurological function (Briner et al. 1999).
Other complications occur in <10% of the patients (aspiration, infection and meningitis) with possibly higher rates for a CSF leak (11–14%) (Moe et al. 1999, Al-Mefty & Teixeira 2002, Lee et al. 2002, Gottfried et al. 2004, Kollert et al. 2006, Suarez et al. 2013a). In contrast to the risks for CBT surgery, the vascular risks are less significant, as these tumours mostly do not invade the carotids; the overall stroke rate is 1.6% (Gottfried et al. 2004). A rare but debilitating complication of JTPGL or VPGL surgery is the ‘first bite syndrome’, severe pain at the beginning of meal consumption (Netterville et al. 1998, Obholzer et al. 2011).
Tympanic PGLs are rarely reported separately; the results are generally more favourable, perhaps because they cause symptoms early and are diagnosed in less advanced stages. The GTR rate is 95–100% with <8% of the cases with non-severe complications and a significantly lower duration of operation and hospital stay. Hearing is generally is maintained (Jackson et al. 1990, O'Leary et al. 1991).
Radiotherapeutic options include conventional fractionated EBRT and SRS.
EBRT has been largely studied for the treatment of HNPGLs and several advantages over surgery have been outlined. Partial/complete symptom relief in the months after treatment occurs in most (52–100%) cases affected by tinnitus, dizziness/vertigo or pain. As opposed to surgery, a CN deficit present at diagnosis generally improves or remains stable (Cummings et al. 1984, de Jong et al. 1995, Huy et al. 2009).
Tumour control with radiotherapy is uniformly defined as a lack of tumour progression (although such an efficacy criterion in tumours with a natural slow growth is debatable); complete tumour remission is very rare, but a slow reduction of tumour volume occurs frequently. Local control occurs in 88–100% of the cases with variable follow-up durations (50 months–11 years). The control rate decreases significantly with time: 95–96% at 5 years and 88–94% at 10 years, but only 73% at 25 years. Disease-related mortality after radiotherapy is low (0–5.1%; Cummings et al. 1984, Hansen & Thomsen 1988, Powell et al. 1992, Verniers et al. 1992, Hinerman et al. 2001, Chino et al. 2009, Huy et al. 2009, Lightowlers et al. 2010, Suarez et al. 2013a).
Mild complications (mucositis, nausea, xerostomia and otitis media/externa) occur occasionally, but are of limited significance (Cummings et al. 1984, Verniers et al. 1992, Hinerman et al. 2008). The most important concerns, especially in young patients, are those regarding serious late effects. Brain or bone necroses are serious adverse effects, albeit being rare nowadays (0.8 and 2.6% respectively; Suarez et al. 2013a). A literature analysis carried out in 1984 has reported brain necrosis to be present in 1.44% of the cases, all receiving higher doses (63–75 Gy; Cummings et al. 1984). Bone necrosis is also correlated with massive radiation doses or concurrent infection (Sharma et al. 1984). Currently, the usual dose is 45 Gy (Hinerman et al. 2008, Huy et al. 2009), and larger doses are no longer used, except for the treatment of malignant tumours, where the response is very poor even then (Hinerman et al. 2008, Moskovic et al. 2010). The radiation-induced malignancy rate is difficult to assess, due to the variable follow-up durations and inconsistency of its ascertainment. Aggressive bone osteosarcoma, fibrosarcoma (Lalwani et al. 1993, Mumber & Greven 1995) and laryngeal carcinoma have been reported up to 25 years after treatment (Lack et al. 1977).
SRS has obvious advantages over EBRT (a single outpatient procedure and better anatomical targeting) and has been advocated as an alternative to surgery. Tumour control is achieved in larger series in 100% of the cases with 31–50% of the tumours slowly decreasing in size (decrease being variably defined). However, the median follow-up duration is short (2–4.8 years), so long-term true local control cannot be accurately assessed. Symptomatic improvement is frequent (29–70%), and a new CN deficit or worsening of a pre-existing CN deficit occurs rarely (0–15%) (Liscak et al. 1999, Foote et al. 2002, Pollock 2004, Lieberson et al. 2012). In a recent literature review, the overall control rate has been reported to be 98% (mean follow-up duration of only 31 months) with a 3% overall complication rate (mostly mild: nausea, vomiting and vertigo) (Lieberson et al. 2012).
In skull base tumours (the most surgically challenging), SRS efficacy has also been reported favourably. A review of the literature published between 1994 and 2004 and a very recent one have reported similar results: a 32–37% decrease in size, 61% remaining stable over an average follow-up duration of 39–41 months. Neurological improvement occurs in 24–39% of the cases, and a few cases experience worsening of conditions (only 2.8% permanent; Gottfried et al. 2004, Suarez et al. 2013a). However, new hearing loss is reported in up to 19% of the SRS-treated JPGL cases (probably due to bone radiation injury; Pollock 2004). JTPGL tumour control with SRS or EBRT is not statistically different, but the mortality due to disease or treatment has been reported to be significantly lower for SRS (Suarez et al. 2013a).
In a meta-analysis of treatment results for JPGLs, the tumour control rate for SRS with a long follow-up duration (mean 71 months) has been reported to be 95%, better than that of GTR (86% at a mean duration of 88 months) (Ivan et al. 2011). This led the authors to conclude that SRS offers better control with a lower risk and should be preferred, reserving surgery for the debulking of larger tumours in preparation for SRS (STR followed by SRS in the same analysis controlled only 71% of the cases, probably due to selection bias). Despite these favourable comparisons, no robust evidence of superiority (e.g. in randomised control studies) exists. Follow-up after SRS is relatively short, so long-term control may be overestimated and the complication rate underestimated. As HNPGLs are generally slow growing (over 90% grow by <0.5 cm/year; Langerman et al. 2012), a definitive assessment of long-term efficacy needs prolonged observation. In addition, the apparent lack of serious complications (e.g. secondary malignancies) needs long-term confirmation. Another limitation is that SRS can generally only be used successfully for the treatment of relatively small tumours (Foote et al. 2002, Chino et al. 2009), i.e. the most likely to be cured surgically without complications.
Treatment of malignant HNPGLs
Data from series of malignant HNPGLs are scarce. The outcome appears to be better with surgery and radiotherapy vs surgery alone. Younger patients tend to respond better, and the overall 5-year survival rate with complex therapy (surgery/radiotherapy/chemotherapy) may become 84% in young (<40 years) patients (Lee et al. 2002, Moskovic et al. 2010).
The results of ‘standard’ chemotherapy are modest (Massey & Wallner 1992, Patel et al. 1995, Pipas & Krywicki 2000, Moskovic et al. 2010), in line with theoretical arguments against the efficacy of classic chemotherapy for slow-growing tumours. Somatostatin analogue (SSA) treatment for receptor-positive tumours has also not shown a major benefit (Duet et al. 2005). The use of modern anti-angiogenic therapies holds some promise, as PGLs are highly vascular tumours; sunitinib (a tyrosine kinase inhibitor) has shown some positive benefits in PGLs not based in the head and neck (Joshua et al. 2009).
PRRT with radiolabelled agents (MIBG or SSA) is an option for the treatment of malignant or inoperable PGLs with high uptake for the specific radiopharmaceutical – up to half of the cases show symptomatic and partial tumour response (Mukherjee et al. 2001, Van Essen et al. 2006, Gonias et al. 2009, Zovato et al. 2012). No studies have evaluated the possible efficacy of PRRT in earlier stages, on a more limited malignant tumour load (perhaps on lower doses that might decrease the risk of severe toxic reactions).
Choice of therapy
For the majority of tumours, complete cure is the most desirable outcome, and surgery is most likely to produce this outcome. However, for tumours with a more indolent natural history, it is important to minimise post-treatment morbidity, and thus therapies offering local control with fewer adverse effects, such as radiotherapy, can sometimes be preferable. Robust evidence in favour of any one treatment method cannot be easily obtained, as randomised trials are rare, current management strategies involve a selection bias, and success criteria have been very differently defined for surgery vs radiotherapy. Overall survival is similar to that of the general population (de Flines et al. 2011), albeit with reduced quality of life (Havekes et al. 2008), so this is not a useful end point for comparison. Therefore, the optimal management of HNPGLs has generated considerable debate over the years, and recommendations have varied from the routine use of radiotherapy (Verniers et al. 1992, Cole & Beiler 1994) to simple observation for most patients (van der Mey et al. 1992).
In the light of current evidence, optimal management is highly dependent on the tumour (location, size, involvement of neurovascular structures, malignancy and hormone production), the patient (age, co-morbidities and symptoms) and the genetic status (implying potential for recurrence, malignancy or multicentric tumours). Pre-treatment assessment should include, where possible, a comprehensive FH, detailed anatomical and functional imaging (to assess location, bone and vascular involvement, multicentricity, functional status, association with other tumours and distant metastases) and a genetic analysis (mutation carriers are younger, with higher rates of multicentricity and malignancy) (Burnichon et al. 2009), as well as an assessment of co-morbidities.
Surgery has been recommended as the treatment of choice in most patients, especially if performed at tertiary centres with surgeons having high expertise and the availability of a multidisciplinary team. As has been noted above, most CBTs (Sajid et al. 2007, Makeieff et al. 2008), small, lower-cervical VPGLs (Urquhart et al. 1994), and class A and B temporal bone tumours (Moe et al. 1999, Suarez et al. 2007) can be cured surgically with an acceptable morbidity rate. Surgery is also the treatment of choice for all catecholamine-secreting PGLs (Young 2006). Advanced or skull base-located tumours are associated with the highest rate of complications and functional disability (with a low response to rehabilitation measures in the elderly). In such tumours, surgery is less effective and more likely to inflict serious damage (Moe et al. 1999, Pareschi et al. 2003). However, many complications are not permanent or do not significantly affect long-term function significantly (Briner et al. 1999). If an extensive CN deficit is present preoperatively, a radical procedure carries a low risk of additional neurological morbidity and offers a high chance of cure, even in advanced tumours (Al-Mefty & Teixeira 2002). In younger (i.e. most) patients, rehabilitation for a surgery-inflicted neurological deficit is more likely to be reversible.
Primary radiotherapy has also been recommended for skull base tumours (Jackson et al. 1990), and recent evidence suggests that it might also be an option for the treatment of CBTs (Suarez et al. 2013b). Indeed, tumour growth arrest following radiotherapy is as frequent as surgical cure, while morbidity is significantly lower (Suarez et al. 2013a,b). Nevertheless, local control decreases with time, and salvage surgery after unsuccessful radiotherapy is technically difficult due to radiation-induced fibrosis. Radiotherapy carries a risk of severe late complications and the possibility of unrecognised malignant potential: irrespective of the type of radiotherapy used, viable cells persist and late distant spread can occur (Chino et al. 2009). In advanced cases, radiotherapy alone implies large-field irradiation with possible deleterious consequences. It would seem reasonable to particularly recommend radiotherapy for older patients, whose risk of late recurrence or complications might exceed life expectancy, and those with bilateral large tumours and/or contraindications to surgery (Moe et al. 1999, Suarez et al. 2007, Evans & Collins 2008, Ma et al. 2009). While all types of radiotherapy may be effective, there appears to be an advantage for radiosurgery where this is technically possible.
For difficult surgical cases, STR followed by observation or radiotherapy is a reasonable approach, as debulking offers symptomatic improvement with a low morbidity rate (Cosetti et al. 2008) and facilitates the use of SRS on a smaller remnant. This conservative strategy has been reported to be successful in patients over 60 years of age (Cosetti et al. 2008), in those with advanced tumours (Moe et al. 1999) or in individuals with previously normal CN function (Tran Ba et al. 2001). However, ‘intended’ STR is rarely described in the literature (frequently STR is the suboptimal result of an attempted total removal), so the theoretical benefit cannot be adequately assessed. Routine adjuvant radiotherapy has not been proven to improve the control rate (Ivan et al. 2011), so radiotherapy should probably be reserved for the treatment of postoperative remnant growth.
But is any intervention always required? Observation alone has been recommended, especially in the elderly and/or asymptomatic patients. (Lieberson et al. 2012). There are studies that have been carried out in The Netherlands showing overall normal life expectancy in HNPGL patients and no survival benefit irrespective of treatment (van der Mey et al. 1992, de Flines et al. 2011): these have been often cited to support observation as a major management strategy. However, simple observation may be inappropriate; if the tumour has malignant potential, unpredictable tumour-associated morbidity may occur, and if the tumour progresses, there may be a need for a later, more hazardous intervention. Watchful waiting can be used in patients deemed to have a short life expectancy (due to age or serious co-morbidities), but it may not be ideal for smaller tumours that may be readily resectable and that may progress or become invasive.
Multicentricity needs to be assessed as it increases the risk of debilitating bilateral neurological deficit (postoperatively and/or by mass effect). Unilateral surgery is usually performed (on the smallest tumour or the side with less CN palsies); contralateral surgery is performed only if no significant CN deficit has occurred. Otherwise, radiotherapy, STR or observation and may be appropriate action only if the tumour becomes symptomatic or grows (Al-Mefty & Teixeira 2002). With genetic screening being more widely implemented, earlier discovery of multifocal tumours and improved treatment outcome are to be anticipated (Fruhmann et al. 2013).
For malignant tumours, surgery and adjuvant radiotherapy are offered for symptomatic relief and improved survival; systemic chemotherapy in unresectable disease and PRRT with radiolabelled agents in high-uptake tumours may be used in selected cases (Patel et al. 1995, Moskovic et al. 2010).
In sporadic cases, annual head/neck MRI for the first 2 years (Papaspyrou et al. 2012) has been proposed. However, the risk of recurrence persists many years after treatment, with a median time to recurrence of 5.8 years (average 8.2) (Jackson et al. 2001); this suggests that a more extended imaging follow-up might be cost beneficial, for at least the first 5 years after treatment. For sporadic CBTs, an annual neck ultrasound initially and then every 5 years has been suggested (Fruhmann et al. 2013), but clinical observation must be maintained indefinitely. Where functional, annual biochemical assessments are required for at least 10 years (Papaspyrou et al. 2012).
In familial cases, the exploration of carriers should begin a decade before the earliest age at diagnosis in the family (Young & Abboud 2006). Annual clinical and biochemical assessments must be performed, with neck ultrasound, CT or MRI being performed every 1–2 years targeting mainly the locations most associated with the pathogenic mutation (Young & Abboud 2006). SSR scintigraphy is also recommended as an adjunct to anatomical imaging for the initial evaluation and follow-up (Gimenez-Roqueplo et al. 2013), but there is no currently validated protocol for long-term follow-up. Where available, DOPA–PET can be used for whole-body screening during follow-up; it should be employed every 2–3 years or to confirm suspicious findings on CT/MRI (Boedeker et al. 2009a,b, Papaspyrou et al. 2012). In any protocol, familial cases should be monitored for multicentric disease indefinitely (Erickson et al. 2001).
The optimal treatment strategy for HNPGLs has not been defined yet. While surgery can be highly effective, it is not uncommonly associated with a high morbidity rate, while radiotherapeutic approaches can prevent tumour progression, but have uncertain long-term consequences. We would in general favour a conservative approach, accepting the need for intervention where there is evidence of tumour progression or where there is concern regarding malignancy. Where surgery is employed, it should generally be confined to smaller tumours in younger patients, while radiotherapy may play a larger role with more invasive or extensive tumours in older patients. A multidisciplinary team is always needed from diagnosis to treatment and follow-up (geneticists, radiologists, vascular surgeons, nuclear medicine specialists, rehabilitation specialists, oncologists, endocrinologists and speech therapists). Further progress in the understanding of PGL pathogenesis is likely to lead to earlier detection, which is essential for successful management.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the review reported.
This review did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.
AstutiDLatifFDallolADahiaPLDouglasFGeorgeESkoldbergFHusebyeESEngCMaherER2001Gene mutations in the succinate dehydrogenase subunit SDHB cause susceptibility to familial pheochromocytoma and to familial paraganglioma. American Journal of Human Genetics6949–54. (doi:10.1086/321282).
BoedekerCCErlicZRichardSKontnyUGimenez-RoqueploAPCasconARobledoMde CamposJMvan NederveenFHde KrijgerRR2009aHead and neck paragangliomas in von Hippel–Lindau disease and multiple endocrine neoplasia type 2. Journal of Clinical Endocrinology and Metabolism941938–1944. (doi:10.1210/jc.2009-0354).
BurnichonNRohmerVAmarLHermanPLeboulleuxSDarrouzetVNiccoliPGaillardDChabrierGChabolleF2009The succinate dehydrogenase genetic testing in a large prospective series of patients with paragangliomas. Journal of Clinical Endocrinology and Metabolism942817–2827. (doi:10.1210/jc.2008-2504).
CasconALopez-JimenezELandaILeskelaSLeandro-GarciaLJMaliszewskaALetonRde la VegaLGarcia-BarcinaMJSanabriaC2009Rationalization of genetic testing in patients with apparently sporadic pheochromocytoma/paraganglioma. Hormone and Metabolic Research41672–675. (doi:10.1055/s-0029-1202814).
CummingsBJBealeFAGarrettPGHarwoodARKeaneTJPayneDGRiderWD1984The treatment of glomus tumors in the temporal bone by megavoltage radiation. Cancer532635–2640. (doi:10.1002/1097-0142(19840615)53:12<2635::AID-CNCR2820531211>3.0.CO;2-9).
van DuinenNSteenvoordenDKemaIPJansenJCVriendsAHBayleyJPSmitJWRomijnJACorssmitEP2010Increased urinary excretion of 3-methoxytyramine in patients with head and neck paragangliomas. Journal of Clinical Endocrinology and Metabolism95209–214. (doi:10.1210/jc.2009-1632).
EisenhoferGGoldsteinDSSullivanPCsakoGBrouwersFMLaiEWAdamsKTPacakK2005Biochemical and clinical manifestations of dopamine-producing paragangliomas: utility of plasma methoxytyramine. Journal of Clinical Endocrinology and Metabolism902068–2075. (doi:10.1210/jc.2004-2025).
Fisch U & Mattox DE 1988 Infratemporal fossa approach type A. Classification of glomus temporale tumours. In Microsurgery of the skull base, pp136-152. Stuttgart, Germany: Thieme
Gimenez-RoqueploAPCaumont-PrimAHouzardCHignetteCHernigouAHalimiPNiccoliPLeboulleuxSAmarLBorson-ChazotF2013Imaging work-up for screening of paraganglioma and pheochromocytoma in SDHx mutation carriers: a multicenter prospective study from the PGL.EVA Investigators. Journal of Clinical Endocrinology and Metabolism98E162–E173. (doi:10.1210/jc.2012-2975).
GoniasSGoldsbyRMatthayKKHawkinsRPriceDHubertyJDamonLLinkerCSznewajsAShiboskiS2009Phase II study of high-dose [131I] metaiodobenzylguanidine therapy for patients with metastatic pheochromocytoma and paraganglioma. Journal of Clinical Oncology274162–4168. (doi:10.1200/JCO.2008.21.3496).
HensenEFSiemersMDJansenJCCorssmitEPRomijnJATopsCMvan der MeyAGDevileePCornelisseCJBayleyJP2011Mutations in SDHD are the major determinants of the clinical characteristics of Dutch head and neck paraganglioma patients. Clinical Endocrinology75650–655. (doi:10.1111/j.1365-2265.2011.04097.x).
JafriMWhitworthJRattenberryEBradleyLKilbyGKumarAVIzattLLallooFBrennanPCookJ2013Evaluation of SDHB, SDHD and VHL gene susceptibility testing in the assessment of individuals with non-syndromic phaeochromocytoma and paraganglioma and head and neck paraganglioma (HNPGL). Clinical Endocrinology78898–906. (doi:10.1111/cen.12074).
JansenJCvan den BergRKuiperAvan der MeyAGZwindermanAHCornelisseCJ2000Estimation of growth rate in patients with head and neck paragangliomas influences the treatment proposal. Cancer882811–2816. (doi:10.1002/1097-0142(20000615)88:12<2811::AID-CNCR21>3.0.CO;2-7).
KahnLB1976Vagal body tumor (nonchromaffin paraganglioma, chemodectoma, and carotid body-like tumor) with cervical node metastasis and familial association: ultrastructural study and review. Cancer382367–2377. (doi:10.1002/1097-0142(197612)38:6<2367::AID-CNCR2820380625>3.0.CO;2-3).
KingKSChenCCAlexopoulosDKWhatleyMAReynoldsJCPatronasNLingAAdamsKTXekoukiPLandoH2011Functional imaging of SDHx-related head and neck paragangliomas: comparison of 18F-fluorodihydroxyphenylalanine, 18F-fluorodopamine, 18F-fluoro-2-deoxy-d-glucose PET, 123I-metaiodobenzylguanidine scintigraphy, and 111In-pentetreotide scintigraphy. Journal of Clinical Endocrinology and Metabolism962779–2785. (doi:10.1210/jc.2011-0333).
LackEECubillaALWoodruffJMFarrHW1977Paragangliomas of the head and neck region: a clinical study of 69 patients. Cancer39397–409. (doi:10.1002/1097-0142(197702)39:2<397::AID-CNCR2820390205>3.0.CO;2-C).
MasseyVWallnerK1992Treatment of metastatic chemodectoma. Cancer69790–792. (doi:10.1002/1097-0142(19920201)69:3<790::AID-CNCR2820690329>3.0.CO;2-U).
MukherjeeJJKaltsasGAIslamNPlowmanPNFoleyRHikmatJBrittonKEJenkinsPJChewSLMonsonJP2001Treatment of metastatic carcinoid tumours, phaeochromocytoma, paraganglioma and medullary carcinoma of the thyroid with (131)I-meta-iodobenzylguanidine [(131)I-mIBG]. Clinical Endocrinology5547–60. (doi:10.1046/j.1365-2265.2001.01309.x).
van NederveenFHGaalJFavierJKorpershoekEOldenburgRAde BruynEMSleddensHFDerkxPRiviereJDannenbergH2009An immunohistochemical procedure to detect patients with paraganglioma and phaeochromocytoma with germline SDHB, SDHC, or SDHD gene mutations: a retrospective and prospective analysis. Lancet Oncology10764–771. (doi:10.1016/S1470-2045(09)70164-0).
NeumannHPErlicZBoedekerCCRybickiLARobledoMHermsenMSchiaviFFalcioniMKwokPBautersC2009Clinical predictors for germline mutations in head and neck paraganglioma patients: cost reduction strategy in genetic diagnostic process as fall-out. Cancer Research693650–3656. (doi:10.1158/0008-5472.CAN-08-4057).
PatelSRWinchesterDJBenjaminRS1995A 15-year experience with chemotherapy of patients with paraganglioma. Cancer761476–1480. (doi:10.1002/1097-0142(19951015)76:8<1476::AID-CNCR2820760827>3.0.CO;2-9).
Rodriguez-CuevasSLopez-GarzaJLabastida-AlmendaroS1998Carotid body tumors in inhabitants of altitudes higher than 2000 meters above sea level. Head & Neck20374–378. (doi:10.1002/(SICI)1097-0347(199808)20:5<374::AID-HED3>3.0.CO;2-V).
SuarezCRodrigoJPMendenhallWMHamoirMSilverCEGregoireVStrojanPNeumannHPObholzerROffergeldCCarotid body paragangliomas: a systematic study of management with surgery and radiotherapyEuropean Archives of Oto-Rhino-Laryngology2013b[in press]doi:10.1007/s00405-013-2384-5).
TimmersHJChenCCCarrasquilloJAWhatleyMLingAHavekesBEisenhoferGMartiniovaLAdamsKTPacakK2009Comparison of 18F-fluoro-l-DOPA, 18F-fluoro-deoxyglucose, and 18F-fluorodopamine PET and 123I-MIBG scintigraphy in the localization of pheochromocytoma and paraganglioma. Journal of Clinical Endocrinology and Metabolism944757–4767. (doi:10.1210/jc.2009-1248).
ZovatoSKumanovaADematteSSansoviniMBodeiLDi SarraDCasagrandaESeveriSAmbrosettiASchiaviF2012Peptide receptor radionuclide therapy (PRRT) with 177Lu-DOTATATE in individuals with neck or mediastinal paraganglioma (PGL). Hormone and Metabolic Research44411–414. (doi:10.1055/s-0032-1311637).