Introduction

Background

Paraganglia are cell clusters located throughout the body in the vascular and neuronal adventitia derived from specialized embryonic neural crest cells [1]. Extra-adrenal paragangliomas (PGs) have been proposed to constitute 15–20 % of paraganglion system tumours [2, 3]. The incidence of all paraganglion system tumours has been estimated by Baysal to be around 1/300,000 [4]. Head and neck paragangliomas (HNPs) are parasympathetic and thus typically non-secreting as opposed to paragangliomas of the lower mediastinum, abdomen or pelvis [5].

Besides being rare, HNPs are slowly growing, highly vascular, and mainly benign tumours that commonly occur at the carotid bifurcation (carotid body tumours, CBTs), at the jugular bulb (jugular PGs, JPs), in the tympanic cavity (tympanic or petrous bone PGs, TPs), or as vagal PGs (VPs) [1, 6, 7]. Malignancy has been reported in less than 10 % of PGs and is determined by metastasis to non-endocrine tissue [3, 7, 8]. Several histological criteria for malignancy have been proposed, but none of these have proven to be conclusive [9].

Genetics

Pheochromocytomas (PCCs) and PGs are frequently associated with an inherited mutation causing susceptibility to paragangliomas [5, 10]. Population studies show that as many as 54 % of patients with seemingly sporadic HNPs have actually a germline mutation in a known susceptibility gene [5]. The risk for genetic background increases with a positive family history of PGs, if the patient is under 40 years, or if the patient also has a PCC or multiple PGs [1, 10]. Presently, ten susceptibility genes have been discovered [5]. The two most common are genes coding for the two succinate dehydrogenase subunits SDHB and SDHD causing paraganglioma syndromes PGL4 and PGL1 [11, 12]. Others include SDHC, SDHAF2, SDHA, TMEM127, MAX and VHL, NF1 and RET (MEN2) [13]. Table 1 outlines the succinate dehydrogenase (SDH) subunit and cofactor genes associated with HNPs. The mode of inheritance for paraganglioma susceptibility is autosomal dominant; however, for SDHD and SDHAF2 this is modified by imprinting and mutations in these two genes cause paragangliomas only when the mutation is inherited from father [14, 15]. Determining the molecular genetic background enables identification of other family members at risk, the mode of inheritance and also predicts the risk of malignant PGs for which the only reliable predictor is a germline mutation in the SDHB gene [16].

Table 1 The SDH susceptibility genes associated with HNPs
Full size table

HNP characteristics

Being neuroendocrine tumours, HNPs are capable of synthesizing a variety of hormonal substances. These, however, are rarely symptomatic unlike their adrenal, abdominal or thoracic counterparts [1, 5]. Primarily the symptoms depend on the location of the tumour, where cervical PGs present with a painless, slowly enlarging lateral neck mass, while tympanic, petrous bone or jugular PGs present with tinnitus and hearing loss as early symptoms [17]. Patients with jugular PGs often suffer from lower cranial nerve deficits as a result of the tumour compression on cranial nerves (CNIX–XI) exiting the jugular foramen [17]. While the average age at diagnosis is approximately 50 years, the sex distribution varies with the site of the tumour; the carotid body PGs are found more often in men (2:1) whereas for vagal PGs the distribution is reversed [1, 18, 19]. The mean tumour doubling rate has been found to be 4.2 years making the growth rate usually slow [6].

Imaging and treatment

HNPs are mostly diagnosed with a computed tomography (CT) or a magnetic resonance imaging (MRI) scan while synchronous carotid body PGs are excluded and carotid body PGs also identified with Doppler sonography [6]. Further functional imaging modalities should be considered such as 18F-dihydroxyphenylalanine positron emission tomography (18F-DOPA PET) if the risk for multiple PGs is high or the patient has an SDHx mutation-related HNP [20]. Fine needle aspiration and biopsy are rarely diagnostic and often risky due to the high vascularity of the tumour [6]. The main treatment of choice for HNPs is still surgical resection, the trend being to move away from radical resection toward function-preserving surgical tumour reduction [18]. Good post-operative results can be expected for CBTs of Shamblin class I and II and TPs, whereas operations on other CBTs and HNPs frequently result in deficits of adjacent cranial nerves [21]. Local control of, e.g. JTPs, can be achieved using post-operative stereotactic radiotherapy (SRT) or conservative wait-and-scan strategy; such function-preserving therapy approach appears to offer the greatest benefit for patients [22]. Tumours’ slow growth, usually benign nature and the tendency of hereditary HNPs to be multifocal may justify less aggressive treatment strategies for all types of HNPs [23, 24].

In this report, we present data on patients diagnosed with histologically verified HNPs in an academic tertiary centre in Finland over a 21-year period. We concentrate on the different management strategies and discuss post-surgery patient outcomes. In addition, we propose a HNP management guideline based on the current knowledge of genetics underlying HNPs (Fig. 1).

Fig. 1
figure 1

Proposal for SDHx-related HNP patient’s and his/her relatives’ medical treatment. 1 If an SDHD mutation is inherited from mother no need for radiological screening, but genetic counselling should be given, because his/hers offspring are at risk. HNP head and neck paraganglioma [1, 5, 16, 20, 40–42, 45]

Full size image

Materials and methods

We retrospectively reviewed the clinicopathological records of all histologically confirmed HNPs managed at the Helsinki University Central Hospital (HUCH), covering approximately 1.8 million inhabitants, in Helsinki, Finland, during a 21-year period between January 1, 1990 and December 31, 2010.

The patients were collected from the archives of the Department of Pathology and their medical, operative and pathologic records reviewed. In total 138 patients were included in the first search from the pathology archives, of which 64 patients filled the inclusion criteria of location in the head and neck area. Two patients were excluded, because their HNP was diagnosed at autopsy. Data on age, sex, symptoms, the timing of diagnosis, tumour location and quantity, embolization, genetic aetiology, Charlson’s Comorbidity Index (CCI), management and follow-up were recorded [25].

Clinical follow-up time was defined as the time period from the time of treatment (i.e. surgery) to the last clinical evaluation or time of death. Overall and disease-specific survival data, consisting of date and cause of death, were provided by the Statistics Finland. Endpoints (overall survival, disease-free survival) were defined according to FDA guidelines [26]. Specifically, when calculating the disease-free survival, non-cancer deaths were censored.

Statistical analysis was carried out using a computerized software package (SPSS, version 19.0, Chicago, IL, USA). Median follow-up time was calculated using the “reverse K-M method” as described by Schemper and Smith [27].

This study was a registry-based study recording the current management of HNP patients during a specific time period with no additional interventions. Therefore, formal Research Ethics Board approval was not necessary. An institutional research permission was granted for the study protocol (§ 78, 27.3.2012).

Results

Patients

We analysed the clinicopathological data of 64 patients with at least one histologically confirmed HNP. There were, in total, 74 HNPs. Six patients had multiple HNPs, which are presented in Table 2. Based on our data the incidence rate was 1.8 new cases per 1,000,000 persons per year. Of the 64 patients, 23 (36 %) were male and 41 (64 %) female with a median age of 56 years at diagnosis (range 21–73 years). At the time of diagnosis, the patients had a median CCI of 0 (range 0–6). Three patients had malignant paraganglioma matching the World Health Organization (WHO) criteria for paraganglioma malignancy [8].

Table 2 Multifocal HNPs listed
Full size table

Presenting symptoms and locations of tumours

The most frequent presenting symptoms were hearing loss or tinnitus in 41 % (n = 26) and a symptomatic neck mass or an incidentally found palpable mass in the neck on physical examination 42 % (n = 27). Two (3 %) patients had high blood pressure as their main symptom while three (5 %) patients were studied for HNP due to positive family history. As a presenting symptom seven patients (11 %) had cranial nerve neuropathies. The cranial nerve deficits were mostly associated with JPs (n = 4), but also CBTs (n = 2) and TPs (n = 1) were involved. 27 (36 %) tumours were located at the carotid body bifurcation (CBT), 19 (26 %) at the jugular foramen (JP), 16 (22 %) in the middle ear/tympanic cavity (TP) while 11 (15 %) were vagal paragangliomas (VP). One tumour was located in the dorsal neck.

Genetic testing

During the clinical workup and treatment of the patients included in our series, genetic testing was not routinely used. Testing was performed only on clinical suspicion as judged by the treating physician. Mutations were identified in four patients as presented in Table 3.

Table 3 Patients with an identified susceptibility gene
Full size table

Preoperative embolization

Preoperative embolization was utilized in 17 operations with HNPs: 50 % (7/14), 50 % (4/8), 22 % (4/18) and 7 % (1/15) of JPs, VPs, CBTs and TPs, respectively. In addition, one dorsal neck PG was also embolized preoperatively. Embolization was used without complications and no difference in post-operative cranial nerve injuries was detected between embolized and non-embolized patients.

Primary surgical therapy

Sixty-two patients representing 68 tumours were treated with primary surgery. Two (3.1 %) patients received only radiotherapy for their HNPs. One patient declined surgery, whereas the other patient’s tumour was considered inoperable. Patients with multiple HNPs received surgical treatment for selected tumours while some tumours were given radiotherapy or the “wait and scan” approach applied (Table 2). The median diameters for operated HNPs were 30 mm (range 20–70 mm) (CBT), 30 mm (range 13–40 mm) (JP), 7 mm (range 3–12 mm) (TP) and 60 mm (range 25–65 mm) (VP) according to the location of the tumour.

Internal carotid artery and cranial nerves were preserved whenever possible. Two of the 25 operated CBTs needed venous grafts to replace the internal carotid artery. A cranial nerve sacrifice was performed in 11 (17.2 %) operations. Nerve sacrifice was deemed necessary in 75 % of VPs (n = 6). In CBTs nerve sacrifice was needed in 11 % (n = 2) and in 50 % (n = 3) of multifocal HNPs. No cranial nerves were excised in JP or TP operations. However, in one TP operation, a controlled breakage of malleus’ handle allowed access to the tumour; this was later repaired.

Follow-up and outcome

Patients were followed without a set guideline. In general, patients had a follow-up appointment at 6 months and 1 year after the operation. The median clinical follow-up time was 4.6 years (range 0–18.8 years). The median follow-up time of patient survival, obtained retrospectively from cause-of-death data, was 11.8 years (range 2.1–22.9 years).

Four (6 %) patients died of the disease. Two of them had malignant tumours and died of tumour progression. Also, one patient suffered a stroke on first post-operative day after surgery. During this procedure, no blood vessel graft was used and the carotid artery was not manipulated, but the internal jugular vein was ligated due to profuse bleeding. One patient developed a fatal myocardial infarction and pulmonary oedema on the third post-operative day. The survival times for patients with malignant HNPs were 15.7–3.6 years. One patient with malignant HNP has survived until the last day of follow-up of 11.9 years. In total 15 patients died during follow-up, 11 of these from causes unrelated to HNPs. The mean overall survival time was 17.8 years and the mean disease-specific survival time 21.3 years. The mean disease-specific survival of malignant cases was 11.5 years, significantly less than the mean disease-specific survival of benign cases at 22.1 years (log-rank test p < 0.001). The two patients that received only radiotherapy were alive at the last day of overall survival follow-up, with follow-up times of 6.8–7.8 years.

Secondary surgery and adjunctive radiotherapy

The mean disease-free time of surgically treated patients was 10.8 years (95 % CI 2.8–18.7), excluding patients with multiple HNPs that received primary radiation therapy to at least one tumour.

Altogether, in 13 (20 %) operations residual tumour tissue was left in situ perioperatively. If tumour tissue was found in post-operative radiological imaging within 6 months of the operation it was also considered to be a residual. Residuals were left in 25 % of patients with a VP, 50 % of patients with a JP, in 13 % of patients with a TP, and in 11 % of patients with a CBT. Residuals were followed up with CT and/or MRI scans and surgery and/or radiotherapy arranged if the growth of the residual tumour was detected.

Nine out of 13 residual tumours were treated. This included both of the malignant cases in our series. Two JP patients received surgery while another two (one malignant CBT and one large TP) received both radiotherapy and surgery. Four patients received radiotherapy only for their residual growth. Four patients were only followed up radiologically. One patient with a malignant residual tumour received adjunctive 131Iodine-metaiodobenzylguanidine (131I-MIBG) therapy after the tumour was re-resected.

During follow-up, four patients had a local recurrence (two JPs, one VP, and one TP). Three of these (two JPs and one VP) were given radiotherapy where the patient with a residual TP received secondary surgery.

Short and long-term post-treatment cranial nerve deficits

The short and long-term cranial nerve deficits included vocal cord paralysis, hoarseness, dysphagia, hearing loss, and Horner’s syndrome. Many sequelae that were recorded at 6 months had recovered by the 12-month follow-up appointment. Table 4 links the post-operative cranial nerve deficits with the location of the tumour. In case of troubling hoarseness vocal cord medialization was performed, typically after at least a year of conservative management.

Table 4 Post-operational sequelae correlated with the location of the tumour
Full size table

Short and long-term post-treatment local complications

Two patients (one JP and one TP) suffered from a post-operative eardrum perforation. Three patients received a tracheostomy at surgery; two of these had multifocal tumours. All patients were decannulated after recovering from the surgery.

Short and long-term post-treatment systemic complications

Aspiration pneumonia, sepsis and meningitis were uncommon in our patient cohort. Two patients had concurrent pneumonia and sepsis; one of these patients also had meningitis secondary to a JP operation.

Discussion

We analysed the management and follow-up data of HNP presenting a diagnostic and therapeutic challenge due to their rarity. Modern imaging (CT and/or MRI), careful clinical examination, and preferably genetic testing are needed before treatment is initiated [28]. Preoperative imaging, genetic tests results and patient’s symptoms should all influence the planned treatment. If tumour removal is possible with minimal morbidity, surgery is the optimal choice [29]. For tumours that cannot be optimally resected or where surgical extirpation poses a high risk of severe and/or multiple cranial nerve damage, conventional radiotherapy or stereotactic radiosurgery should be considered [6, 29]. Surgery of HNPs should be discussed and planned preoperatively in a multidisciplinary environment with adequate rehabilitation possibilities to prevent unnecessary patient morbidity. There is a clear need for more organized and prolonged preoperative assessment, including genetic testing, and follow-up of HNP patients.

The sex distribution and median age in the present study were consistent with other studies [18, 19, 30, 31]. In accordance with previous studies, CBTs were the most common HNPs in our study, their percentage (37 %) being slightly lower than in other reports [31, 32]. CBTs were followed by JPs (26 %), TPs (22 %) and VPs (15 %) with the percentage of TPs and VPs showing higher prevalence than in other studies [31, 32]. The incidence rate of 1.8 cases per 1,000,000 per year is similar as previously reported [4]. Overall, our study presents accurate and statistically confirmed data covering important survival endpoints from an area of nearly 2 million inhabitants.

Boedeker et al. [6] suggest using preoperative embolization in patients with JPs, Shamblin class II and III CBTs and VPs with a diameter larger than three centimetres. In a study of 131 CBTs (104 Shamblin class II or III), Power et al. [33] found that preoperative embolization reduces intraoperative blood loss and may simplify the conduct of the operation, but does not decrease the rate of cranial nerve injuries. Duffis et al. [34] conclude that preoperative embolization is an effective and safe adjuvant to surgical resection when the operator has adequate training, knowledge of anatomy and experience of procedures at hand. In our cohort of patients, embolization was used without complications and no difference in post-operative cranial nerve injuries was detected between embolized and non-embolized patients.

Fifty per cent of the resections on JPs were complete. The large percentage of subtotal resections can be explained by the avoidance of damaging cranial nerves and vasculature in the skull base. The extent of resection is restricted by post-operative sequelae that can be expected as shown in Table 4. In a review article of JPs and VPs by Suárez et al. [23], surgical control was achieved in 85 % of cases (including 14 % with subtotal resection). As highlighted by Suárez et al. [23], the benefits must outweigh the possible morbidity of the operation. A wait-and-scan approach as well as radiotherapy or radiosurgery should be considered for JPs that are subtotally resected or not suitable for surgery, i.e. when complete excision would likely lead to cranial nerve injuries [23]. Suarez et al. [23] suggest that indications for surgery and radiotherapy of JPs and VPs should be more strict and limited to those causing symptoms or being radiologically progressive (volume growth of over 20 % in 1 year). Stereotactic radiosurgery and especially gamma knife radiosurgery should be considered for JPs as these have been shown to dramatically reduce the radiation dose that falls on healthy tissue [35]. Still, even if good short-term control rates have been reported with stereotactic radiosurgery, there are yet no reports on the long-term follow-up data [35]. At our institution stereotactic radiotherapy, but not radiosurgery, is used.

In VP surgery, vagal nerve palsy is often unavoidable resulting in significant morbidity. If a patient has a VP but no cranial nerve deficits, surgery should not be done and instead, “wait and scan” approach is recommended [23]. The results of CBTs in the present study were in agreement with previous studies [36, 37]. CBTs’ appropriateness for surgery has also been discussed by Suarez et al. [24] and surgery should be reserved for sporadic or familial unilateral diseases, where excision does not cause extra morbidity. Bilateral CBTs should be managed with a “wait and scan” approach or at least one tumour left intact to avoid bilateral carotid sinus denervation and the risk of baroreflex failure syndrome [38]. Treating one or both tumours with radiotherapy is another possibility [24]. In the present study, TPs were typically treated without complications and the follow-up was brief.

In our study, a malignant HNP was an adverse indicator for survival as two out of three patients with such a tumour died of the disease. There is no consensus on the optimal treatment of malignant HNPs. Mendenhall et al. [29] have suggested primary surgery followed by post-operative radiotherapy (60–70 Gy). However, detecting malignant forms of HNP is challenging, the metastases tend to grow slowly and they may be present at primary presentation or may not develop clinically apparent until decades later. Furthermore, their histology is not conclusive [3, 9]. Currently, the only known predisposing factor for malignancy is an SDHB mutation [3, 16, 39].

Since many residual HNPs and local recurrences appeared during the follow-up, we recommend a longer follow-up for HNP patients. Local recurrences in the present study were detected 2.5–10.8 years after the primary surgery, while our median follow-up time was 4 years. Based on this, even longer follow-up periods are justified. The genetic testing results should also be considered when determining the length of the follow-up and imaging methods. For patients with inherited paraganglioma susceptibility, the follow-up should be lifelong.

In the present study, there was no protocol for genetic screening of patients with HNPs, because genetic testing was not widely available for SDHx mutations in the 1990s. More importantly, generally accepted guidelines regarding whom to send for testing and genetic counselling are needed. Recent studies have suggested that all HNP patients should be screened for at least SDHD and SDHB and, if needed, SDHC mutations [5, 40, 41]. This should be done not only because it will significantly improve the chances of finding patients’ at-risk relatives early on, but also because many genetic mutations may otherwise go undetected [5, 40, 41]. Finding a gene mutation is also important for patient follow-up, primary investigation and treatment to be planned correctly. Boedeker et al. [13] recommend a lifelong follow-up of SDHB, SDHC and SDHD mutation carriers with MRI of head and neck, thorax and abdomen and if needed with functional imaging.

CT or MRI scanning is advocated for primary anatomical imaging [20]. When choosing functional imaging for SDHx-related HNPs, 18F-DOPA PET possesses the highest accuracy and is therefore the optimal screening method [1, 20, 40]. 18F-DOPA PET also allows the screening of the whole body, which is particularly useful for patients with an SDHB mutation that predisposes to abdominal and thoracic PGs and PCCs [1, 40]. The golden standard for the best functional imaging of PGs of abdomen and thorax and PCCs has, however, not been decided. If strong suspicion for other PG or PCC arises, imaging should be planned with a nuclear medicine physician [42]. Because tumour size correlates strongly with surgical and post-operative complications, it would be advantageous to diagnose HNPs early on [16, 43]. Some residual tumours might only be detectable by angiography [44].

Based on our findings and the past decades development in the understanding of genetic background of HNPs and developments in functional imaging we suggest, that all HNP patients should be screened for SDHB, SDHD and, if needed, SDHC mutations [5, 13]. If these tests are negative, but a strong clinical suspicion remains then other susceptibility genes should be tested for. We have made a proposal for genetic testing and functional imaging pathway for HNP patients and their relatives. The pathway (Fig. 1) is based on the likelihood of SDHx-related mutations playing a major part in causing inherited and sporadic mutations leading to HNPs.

Conclusions

Surgical treatment of HNPs should be properly planned considering the patient’s preoperative state and the expected morbidities following surgery. Patients with multiple HNPs form an especially challenging group where a combination of different treatment options should be considered. A 4-year follow-up of HNP patients is too short for all patients and individual genetic testing results will influence the screening protocol. For all patients with HPGs, family history should be collected and genetic consultation and testing organized, where appropriate. In the near future, if not already, genetic testing panels for patients with PGs, will greatly improve the efficiency of molecular genetic testing, and should be considered for all patients [45]. More follow-up data and studies are still needed regarding the sporadic and hereditary background of local recurrences and new tumours. Based on these results, patients with and without a susceptibility gene mutation need individually designed intervals for imaging and clinical follow-up visits.