Combined percutaneous and transarterial devascularisation of juvenile nasopharyngeal angiofibroma with protection of internal carotid artery: A modification of the technique

Abstract

Juvenile nasal angiofibroma (JNA) is a hypervascularised, benign, but locally aggressive tumour that grows in the posterior, upper part of the nasal cavity and invades surrounding anatomical structures. The treatment of choice is surgical removal, but complete resection of the tumour can be hampered because of profuse perioperative bleeding. Preoperative embolisation of the tumour has been proposed as an effective method for prevention of perioperative bleeding, thereby shortening of the time of the operation. In this report of five cases, we describe successful preoperative devascularisation of the tumour by applying a modified method of direct intratumoural injection of the liquid embolic agent Onyx combined with protection of the internal carotid artery. The control of bleeding during the embolisation and occlusion of the maxillary or sphenopalatine artery was achieved by using a bi-luminal balloon catheter. Such use of the dual-lumen catheter in treatment of JNA has not been reported so far in the medical literature.

Introduction

Juvenile nasal angiofibroma (JNA) is a benign, locally invasive, hypervascularised tumour without a true capsule.¹ This tumour type accounts only for about 0.05% of all head and neck tumours and occurs almost exclusively in teenage males.² The tumour originates from the submucosal tissue of the superior margin of the sphenopalatine foramen, posterior to the middle turbinate, and may expand over time to the pterygopalatine and infratemporal fossa.³ Most patients with advanced JNA present with signs of nasal obstruction and the majority of them with epistaxis.⁴ Surgical resection of the tumour, which is the treatment of choice, is difficult both as an open surgical procedure as well as with use of an endoscopic technique.⁵ According to Ardehali et al. the mean intraoperative blood loss in patients with embolised tumour was 770 ml compared with 1403.6 ml in patients with non-embolised tumours.⁶ Preoperative embolisation of JNA has been performed since 1988 via a transarterial route using particulate embolic material and since 1994 by the direct intratumoural injection of an embolic agent.^7,8 Multiple, small and tortuous feeders as well as the spasm caused by manipulation of the catheter may limit the effects of the embolisation. Moreover, an inadvertent embolisation of vital structures via anastomoses between branches of the external and internal carotid artery with serious clinical consequences such as blindness, stroke and even death may occur during treatment regardless of the route of injection of the embolic material.^6,9,10 In this report we describe our experience with treatment of JNA in five patients using direct intratumoural injection of the embolic material Onyx and transarterial embolisation of the maxillary artery together with simultaneous balloon protection of the internal carotid artery. Embolisation of the maxillary artery was made through dual-lumen balloon catheter which enabled not only rapid embolisation of this artery, but also effective control of bleeding during repeated punctures of this hypervascularised tumour.

Clinical material, embolisation technique and results

We have treated five patients, all males, aged between 14 and 34 years (mean age 24.2 years). All patients were informed of the treatment strategy, and all interventions were carried out in accordance with applicable local and general ethical standards and principals. Written consent for publication of this article was obtained from all patients before the submission of the manuscript.

Two patients presented with nasal obstruction and three with epistaxis. In two patients, the tumour was supplied by the ipsilateral maxillary artery and ipsilateral internal carotid artery, in two cases only by the ipsilateral maxillary artery, and in one case by the ipsilateral internal carotid artery and both the ipsilateral and contralateral maxillary arteries (Table 1).

Table 1.

Demographic, technical and procedure-related data.

Gender Case number		Age (yr)	Symptoms	Supply		Catheter in maxillary artery	Balloon catheter in ICA	Embolic material for maxillary artery	Volume of Onyx 18 injected in JNA	Total fluoroscopy time (minutes)	Complication related to embolisation/ consequences	Size of the tumour (LL × CC × AP)	Blood loss (ml)
				ICA	Maxillary artery
1	m	26	epistaxis	ipsilateral	ipsilateral	Scepter 4 × 10	5 × 15 HyperGlide	Coils	3 ml	49.2	Leakage of the Onyx along inferior surface of the tumour/none	13 mm × 28.9 mm  × 24 mm	200 ml
2	m	32	nasal obstruction	none	ipsilateral	Scepter 4 × 10	5 × 15 HyperGlide	Coils	4 ml	33.6	None	17.7 mm × 40.2 mm  × 28.8 mm	Bleeding not measured/negligible
3	m	34	nasal obstruction	ipsilateral	ipsilateral	Scepter 4 × 11	5 × 20 HyperGlide	Coils	3 ml	63.8	None	14 mm × 13.4 mm  × 21.5 mm	<200 ml
4	m	14	epistaxis	ipsilateral	bilateral	Scepter 4 × 10	4 × 10 HyperGlide	Onyx 18	5 ml	108	Leakage of the Onyx along posterior surface of the tumour/none	37.6 mm × 39.2 mm  × 37.6 mm	500 ml
5	m	15	epistaxis	none	ipsilateral	Scepter 4 × 10	5 × 20 HyperGlide	Coils	4 ml	35.2	None	24.3 mm × 13 mm  × 28.6 mm	300 ml

The interventions were performed with the patients under general anaesthesia. In all cases a 6 F Envoy guide catheter (Cordis Corp., Miami Lakes, FL, USA) introduced via a sheath inserted into the right femoral artery was placed distally in the common carotid artery. In cases with bilateral arterial supply, guiding catheters were placed in both common carotid arteries. After placement of the guide catheter/catheters, a bolus of 3000 units of Heparin was administered intravenously for prevention of thromboembolic complication. A compliant HyperGlide balloon (Micro Therapeutics, Inc d/b/a ev3 Neurovascular, Irvine, CA, USA), of size depending on the size of internal carotid artery (Table 1), was introduced through the guiding catheter and placed in the cavernous segment of the internal carotid artery. In the cases where the tumour was supplied by the mandibulovidian artery, the balloon was placed in the petrous segment of the internal carotid artery. A Scepter 4 × 10 or 11 dual-lumen balloon catheter (MicroVention, Inc. Tustin, CA, USA) was introduced through the same guiding catheter and placed distally in the internal maxillary artery. A rigid rhinoscope was then introduced between the inferior and middle turbinate. A late parenchymal phase of the angiography of the external carotid artery was used to obtain a road map in the lateral projection. The tumour was then punctured by a 0.9 × 150 mm 20G needle (Mediplast AB, Malmö, Sweden) under simultaneous fluoroscopic and endoscopic guidance (Figure 1). In one case, the needle was bent to enable puncture and embolisation of the lateral part of the tumour (case no. 4, Figure 2). Correct positioning of the needle was confirmed by slow and continuous blood reflux. Contrast was injected to obtain a parenchymogram in a lateral projection. The dimethyl sulfoxide (DMSO) and Onyx syringes were mounted directly onto the needle, without a connecting tube. The dead space of the needle was filled with DMSO (Micro Therapeutics, Inc d/b/a ev3 Neurovascular, Irvine, CA, USA). The tumour was then embolised by slow injection of an ethylene vinyl alcohol copolymer in the form of a liquid embolic agent (Onyx 18, Micro Therapeutics, Inc d/b/a ev3 Neurovascular, Irvine, CA, USA) under continuous fluoroscopic control. The exchange of syringes was performed quickly to prevent reflux of blood into the needle and thus premature polymerisation of Onyx in the needle. The total amounts of Onyx used in these embolisations, as given in Table 1, are not absolutely reliable because minor volumes of the embolic agent were lost during the exchanges of the syringes. If the tumour had to be punctured several times because of only partial penetration of the embolic agent, bleeding from the puncture site was prevented or immediately stopped by temporarily inflating the double-lumen balloon placed in the maxillary artery. This manoeuvre enabled rapid and clean re-punctures of the tumour and was effective even in cases with double supply because the contribution of the mandibulovidian artery to the supply of the tumour was marginal.

Figure 1.

Case number 1: right external carotid angiography shows the vascular bed of the tumour composed of numerous densely packed tiny blood vessels and supplied mainly by the internal maxillary artery (a); the tumour is also fed by mandibulovidian artery originating from the petrous segment of the internal carotid artery (arrow) (b); non-subtracted lateral X-ray of the facial region shows non-detachable balloon placed in the internal carotid artery (arrow head), dual-lumen balloon placed in the maxillary artery (long arrow), tip of the needle (short arrow) in the tumour – note the contrast injected in the core of the tumour and rhinoscope inserted between middle and lower turbinate (parallel arrows) (c); non-subtracted lateral X-ray of the facial region shows the Onyx cast in the tumour and catheters still in arteries (d); non-subtracted AP X-ray of the facial region shows Onyx cast in the tumour (long arrow), coils in the maxillary artery and dual-lumen catheter (short arrow) (e); post-interventional right common carotid artery angiography shows no contrast in the tumour (f).

Figure 2.

Case number 4: left maxillary artery angiography shows the vascular bed of the tumour extending from the right nasal cavity, crossing the midline and reaching the left pterygopalatine fossa (a); AP road map showing the bent needle (long arrow), inflated balloon of the dual-lumen balloon catheter (short arrow) and non-detachable balloon in the left internal carotid artery (arrow head) (b); non-subtracted AP X-ray of the facial region shows Onyx cast in the tumour and the leak of the Onyx along posterior surface of the tumour and soft palate (arrow) (c); left maxillary artery angiography shows no contrast in the tumour (d).

Repeated angiographies of the external carotid artery were performed to determine the size of the non-embolised part of the tumour. After the embolisation of the tumour was completed, the internal maxillary artery was occluded either by coils or, in one case, by Onyx (Table 1). The balloon placed in the internal carotid artery was not used because Onyx filled only the pathological vascular structures within the tumour. In one case, Onyx penetrated the posterior and in one the inferior surface of the tumour during the injection (cases 1 and 4). In case 1, the Onyx cast was aspirated easily by the rhinoscope, whereas in case 4, the Onyx cast remained firmly attached to the surface without causing any clinical consequences (Figure 2).

The mean fluoroscopy time for three patients (cases 1, 2 and 5) was 39.3 minutes. This fluoroscopy time includes preoperative angiographic work-up, injection of Onyx, embolisation of the internal maxillary or sphenopalatine artery as well as the final angiographic control. This time was significantly exceeded in case 3 due to anatomical reasons. The space between the nasal turbinates was narrow and thus limited the approach to the tumour, and the lateral extension of the tumour was pronounced. The time was also exceeded in the case 4 because of the enormous size and extension of the tumour. In this patient, the tumour occupied the space between the right nasal cavity and the left pterygopalatyne fossa in the coronal plane and between the central part of the left nasal cavity and left sphenoid sinus in the sagittal plane.

All patients were neurologically intact after the intervention. The tumour was removed in all cases within 24 h after the embolisation. The mean total blood loss was less than 300 ml in our series.

All patients underwent surgery through an open approach, and in the majority of the cases, a significant part of the blood loss during surgery was due to the approach per se rather than the removal of the tumour.

Discussion

JNA is a benign, locally invasive tumour that originates from submucosal tissue in proximity to the sphenopalatine foramen and the middle nasal turbinate. It may invade surrounding extra- and intracranial spaces by eroding the skeletal boundaries or by spreading via anatomical openings and channels. Patients present with a variety of symptoms depending on the degree of expansion of the tumour.^3,4,11 Surgical removal of JNA has been regarded a challenge for the surgeon because of profuse intraoperative bleeding from both the tumour and its attachments. That is why preoperative devascularisation of the tumour either via the endovascular route or by direct intratumoural injection of the embolic material has become an important supplementary tool in the treatment of JNA.^7,8 In addition, a successful preoperative embolisation of the tumour contributes to better exposure of the tumour, shorter time of surgery and a more radical removal of the tumour.

Particulate embolic material injected via a transarterial route has been used for the devascularisation of JNA since the 1980s.^7,12 The most important disadvantage with particulate embolic agents is poor visibility. The PVA particles suspended in the diluted contrast are not roentgen-opaque, and it is very difficult to trace the diluted contrast medium in the tiny distal branches of the internal maxillary artery and collaterals between branches of the external and internal carotid arteries.¹³ There are reports of severe complications caused by inadvertent embolisation of the orbital or cerebral arteries by particulate embolic agents.^9,14

Another important disadvantage of transarterial embolisation is only partial devascularisation of the tumour caused by the unfavourable anatomy of numerous and tortuous feeding arteries as well as the arterial spasm which sometimes occurs after prolonged catheterisation of branches of the external carotid artery.¹⁵ In addition, transarterial embolisation is almost always associated with a relatively long fluoroscopic time, which often exceeds 50 minutes.¹⁶ These disadvantages have been overcome by direct injection of embolic material via a needle placed in the tumour under fluoroscopic and/or endoscopic guiding. This method has been in use since 1994.⁸ The injection time is shorter, the infiltration of the tumour is deeper and intraoperative blood loss less.¹⁷

The first liquid embolic agent that was used for preoperative devascularisation of JNA was N-butyl-cyanoacrilat (NBCA).⁸ This embolic agent can be injected either via the transarterial route or by direct puncture of the tumour. NBCA is roentgen-opaque and can be traced easily in the vasculature. The main disadvantage of NBCA is a rapid polymerisation, and consequently, if injected via the transarterial route, proximal embolisation of the vascular bed of the tumour may occur. Moreover, inadvertent embolisation of branches of the internal carotid artery by the NBCA regardless of the route of injection has been reported.^10,18,19 In order to prevent these devastating complications, Ming et al. punctured the tumour under local anaesthesia and then injected 2% lidocain (1–2 ml) through the needle into the tumour. The patients were then monitored for 5 min. If signs of cranial nerve dysfunction occurred, the needle was immediately repositioned. The NBCA was injected when needle position for safe injection of the NBCA was ascertained.²⁰

Ethylene vinyl alcohol copolymer (Onyx) dissolved in DMSO has long been used for devascularisation of numerous hypervascularised lesions in the head and neck region as well as in the endocranium.²¹ Onyx can be injected directly into the tumour or by a transarterial route. If Onyx is injected by a transarterial route, only proximal occlusion of some feeding arteries can be achieved. A nearly complete devascularisation of the tumour can be achieved only if the Onyx is injected directly into the tumour tissue.¹⁶ Lehman et al. reported direct intratumoural injection of Onyx with simultaneous protection of the internal carotid artery by a non-detachable balloon, which contributed to a safer devascularisation of the tumour.²² Onyx, as a liquid embolic agent, has numerous advantages compared with particulate embolic agents and NBCA. The visibility of Onyx is significantly better than the visibility of other embolic agents thanks to tantalum powder added to the ethylene vinyl alcohol copolymer. The polymerisation of Onyx is not as rapid as polymerisation of NBCA, and Onyx fills the vascular spaces very slowly, which makes embolisation by Onyx much more effective and controllable.^17,22,23 This feature of Onyx also enables control runs during the embolisation, which are necessary in order to estimate the size of the non-embolised part of the tumour. In three of the five cases in our series, the tumour was fed not only by terminal branches of the external carotid artery, but also by tiny extradural branches of the internal carotid artery.

Inadvertent migration of the PVA micro-particles or NBCA into the cerebral arteries is a serious potential complication of preoperative embolisation of JNA.^{9,10,14,18,19} Ballah et al. recently published a series of 17 paediatric patients who were treated by transarterial injection of PVA micro-particles. Almost complete devascularisation was achieved in all cases without any procedure-related complication. However, the mean preoperative blood loss exceeded 600 ml and the mean total angiography and embolisation procedure time was 134 min. The mean fluoroscopic time was 36 min, but this figure was obtained from only six available procedures.²⁴

To date, inadvertent Onyx migration into the cerebral arteries during embolisation of tumours in the head and neck region has not been reported. Nevertheless, the authors decided to prevent this potentially very serious complication by placing a non-detachable balloon in the internal carotid artery.^22,23

Two important steps in the surgical treatment of JNA are detection and ligation or cauterisation of the sphenopalatine artery, which are performed to prevent profuse bleeding during removal of the tumour.⁴ Preoperative occlusion of this artery may make this phase of surgery unnecessary and can thus shorten the operative time, increase the safety of the operation and contribute to a decrease in the total perioperative blood loss. In those cases in our series of patients in which ligation of the sphenopalatine artery was necessary to ensure complete tumour removal, this was relatively easy due to limited intratumoural bleeding. In all cases in our series, the dual-lumen Scepter balloon was used. This balloon is compliant, very flexible and visible, has good navigability and can be safely used even for the embolisation of cerebral AVMs fed by distal branches of the main cerebral arteries.^25,26 Perfect temporary control of bleeding was achieved by inflating this balloon in the distal maxillary artery during the repeated punctures of the tumour. In four cases, the distal maxillary artery was occluded by coils and in one case by Onyx. In all these cases, the balloon was inflated during embolisation of the artery, which thus prevented a proximal migration of the coils as well as the reflux of Onyx along the catheter.

The mean total blood loss in our series was less than 300 ml, which is significantly less than reported by others. According to the literature, the total blood loss depends on the stage of the tumour and treatment strategy, and varies between 770 ml (in embolised cases) and 1500 ml in (non-embolised cases).^6,27

In all cases, surgery was performed through an open approach, which is the tradition at our centre. In several cases, a significant part of the perioperative blood loss was due to the approach itself rather than the tumour removal. However, with a reliable preoperative embolisation such as described here, an endoscopic approach would be feasible, at least for the less advanced cases and even for the less than superspecialised endoscopist. In addition to reduced blood loss, an endoscopic approach would also result in reduced scarring of the face.

The mean fluoroscopy time in our three non-complicated cases was 39.3 min compared with 50 min, which is the mean fluoroscopy time in cases treated with transarterial injection of embolic material.¹⁶

Conclusion

To improve the technique of preoperative embolisation of JNA, the authors combined all the successful approaches. The tumour was punctured by a needle under combined endoscopic and fluoroscopic guidance, which guaranteed a proper placement of the needle. The shape of the needle was adapted to the size, shape and localisation of the tumour.

Control of the infiltration of the tumour and migration of the embolic agent were obtained by continuous fluoroscopy. A non-detachable balloon placed in the cavernous segment of the internal carotid artery protected the orbital and cerebral arteries from the potential, inadvertent migration of Onyx into the cerebral arteries. Finally, the use of the dual-lumen balloon catheter enabled excellent control of bleeding during the repeated punctures of the tumour as well as very rapid and effective occlusion of the distal maxillary or sphenopalatine arteries. Such use of the double-lumen balloon catheter in the treatment of JNA has not been reported so far the medical literature.

Using this modified and safe technique we achieved a significant shortening of the time for both preoperative embolisation and surgical removal of the tumour as well as a significant reduction of total perioperative blood loss.

Ethical considerations

All patients were informed of the treatment strategy, and all interventions were carried out in accordance with applicable local and general ethical standards and principals. A written consent for publication of this article was obtained from all patients before the submission of the manuscript.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest

The authors declare no conflict of interest.