Erosion of the nostril at the end of surgery (with absorbable nasal packing).
Patient with stage IIIA tumor. Endoscopic access to the pterygopalatine fossa with a 30°-angled telescope (black lines). Maximum amplitude of axis position with a 0° telescope (white lines).
Patient with stage IIC tumor. Preoperative computed tomographic scan: axial (A and B) and coronal (C and D) views. E, Magnetic resonance imaging scan, coronal view. Note the erosion of the pterygoid plates and sphenoid bone (A and D, arrows).
Same patient as in Figure 3. Postoperative magnetic resonance imaging scan at 9 months (A and B) and 18 months (C and D). Small asymptomatic remnant (A-D, arrows), nonevolving with time.
Patient with stage IIIA tumor. Magnetic resonance image scan: axial (A and B) and coronal (C and D) views. Note adhesion to the infraorbital fissure (C and D, arrows). E, Sagittal view. F, En bloc resection.
Same patient as in Figure 5. Postoperative computed tomographic scan (with iodine injection) at 12 months: axial (A-C) and coronal (D-F) views. Note the mucosal inflammation without tumor recurrence.
Roger G, Tran Ba Huy P, Froehlich P, Van Den Abbeele T, Klossek J, Serrano E, Garabedian E, Herman P. Exclusively Endoscopic Removal of Juvenile Nasopharyngeal AngiofibromaTrends and Limits. Arch Otolaryngol Head Neck Surg. 2002;128(8):928-935. doi:10.1001/archotol.128.8.928
To determine the feasible conditions for exclusive endoscopic resection of juvenile nasopharyngeal angiofibroma.
Retrospective study of 20 patients, with a mean follow-up of 22 months.
Six academic referral hospitals.
All patients had a preoperative computed tomographic or magnetic resonance imaging scan and at least 1 follow-up computed tomographic and/or magnetic resonance imaging scan 6 or 12 months after surgery. Exclusive endoscopic removal was performed using conventional functional endoscopic sinus surgery instrumentation after preoperative embolization.
Using Radkowski staging, 4, 7, and 9 patients had stage I, II and IIIA tumors, respectively. Seven patients were operated on for a recurrence after open surgery. Extension toward the sphenoid sinus, pterygomaxillary fossa, or infratemporal fossa could be removed. There was no attempt at endoscopic removal of deep skull base or temporal fossa invasion. The mean surgery duration was 135 minutes; mean dimensions of the tumor were 4.5 × 4 × 3 cm; and mean blood loss was 350 mL (median, 300 mL). No recurrences occurred in this series; there were small asymptomatic remnants in 2 cases.
An exclusively endoscopic management of juvenile nasopharyngeal angiofibroma appears to be effective for small to medium tumors. It should be considered as a first-choice option for these cases (in view of the minimal bleeding, shorter duration, and efficacy).
JUVENILE NASOPHARYNGEAL angiofibroma (JNA) is a relatively rare, benign vascular tumor that arises at the sphenopalatine foramen1 and then extends to the nasal cavities and the nasopharynx to eventually reach, and even extend beyond, the base of the skull by posterior or lateral invasion. The latter localization is the principal site of recurrence.2,3 However, the spontaneous involution of a postsurgery remnant is well documented in the literature.4,5
There are a number of therapeutic options, although the most common initial treatment is surgical excision preceded by embolization.1- 3,6,7 The various surgical approaches that have been used range from transpalatine to a lateral rhinotomy and to midfacial degloving,8 while for some groups a neurosurgical approach is also required. The use of endoscopic surgery for the resection of JNA has only recently been reported,9 and there are fewer than 50 cases documented in the literature.2,7,10- 20 Most of these patients had relatively limited tumors, Radkowski stage IIA at most.3 Nevertheless, in an article by Jorissen et al,18 7 of 13 cases were more advanced than stage IIA. Apart from this group, significant involvement of the pterygopalatine or infratemporal fossae has been regarded by most surgeons as a contraindication to a purely endoscopic approach to sinus surgery. On the other hand, Carrau et al20 considered that involvement of the pterygopalatine or infratemporal fossae could be treated by endoscopy alone. With respect to the use of endoscopic surgery for the excision of maxillofacial tumors,21,22 it seemed appropriate to us to reevaluate the limits of endoscopic resection of JNA.
The medical notes of patients operated on for JNA (confirmed histologically) by an exclusively endoscopic approach between 1995 and 2000 were reviewed in 6 centers. Ten patients were operated on at Lariboisière Hospital (Paris, France), 5 at Armand-Trousseau Children's Hospital (Paris), 2 at Edouard Herriot Hospital (Lyon, France), 1 at Robert Debré Children's Hospital (Paris), 1 at Rangueil Hospital (Toulouse, France), and 1 at Jean Bernard Hospital (Poitiers, France).
Seven patients had relapses of tumors that had been operated on by a transfacial approach (mean time, 18 months prior). All were male patients aged between 11 and 32 years (median age, 15 years). The main presenting symptom was nasal obstruction, which was present in 17 patients, and in 2 cases the nasal obstruction was associated with a genuine obstructive sleep apnea syndrome. Fifteen patients also had epistaxis of variable severity. In addition, 2 patients had serous otitis media, and 1 had hypoesthesia of the infraorbital nerve.
All 20 patients had a preoperative computed tomographic and/or magnetic resonance imaging scan, as well as preoperative arteriography with embolization (except for 1 patient). In addition, all underwent a follow-up computed tomographic and/or magnetic resonance imaging scan at least 6 months after surgery for 5 patients and greater than 12 months for the 15 others (range, 12-60 months; median, 18 months). The techniques used for evaluating the extent of invasion by the tumors have been reported previously.7 Eighteen different topographic sites were systematically analyzed, and the results are given in Table 1. Arteriography with embolization was performed systematically with the exception of 1 patient who had a relapse limited to the sphenoid sinus. No complications occurred with the 19 arteriographies. The embolization technique has been previously reported.23,24 The branches of the external carotid artery were systematically embolized. In 2 cases, the branches of the contralateral external carotid also needed to be embolized. Embolization of the tumoral vasculature arising from the internal carotid artery was performed in 6 cases, preceded by a test occlusion. Embolization by direct injection was also necessary in 2 cases. The devascularization was judged to be complete in all but 2 cases (a residual blush of 10% to 20% was related to inaccessible vessels arising from the internal carotid artery). The surgical resection was performed between 1 and 6 days after embolization (median delay of 2 days).
All patients were staged using the Radkowski classification,3 derived from the Sessions classification (compare with Table 1),25 which, within stage III, allows differentiation from an erosion of the base of the skull, corresponding to stage IIIA.
A protocol of autologous blood donation was systematically used to collect 1 to 3 U of packed red cells (depending on the hemoglobin level at the time of diagnosis) during the 3 weeks prior to surgery. Surgery was preceded by arteriography with embolization. Surgery was performed under general anesthesia. The nasal fossa was packed with cotton swabs soaked in a vasoconstrictor and left in place for 15 to 20 minutes. The surgery was performed using a 30° telescope in all cases, as well as a 0° telescope in 1 case. Standard surgical endonasal instrumentation was used, often accompanied by an Obwegeser raspatory. Usually, the first point of intervention was the septum where the tumor was frequently attached and where the vasculature arising from the posterior ethmoidal artery (a branch of the ophthalmic artery) had not been embolized. A bipolar coagulation of the septal mucosa all around the zone of insertion of the tumor was performed before attempting to release the septal attachment. When the vomer was deeply involved, the surgery started with sectioning of the septum just in front of the tumor. The tumor was then progressively detached by pushing it downward and posteriorly, which exposed the choanal arch, enabling a sphenoidotomy around the tumor. Excision of tumor in the sphenoid sinus could then usually be completed by progressive traction and detachment. Thereafter, the region of the sphenopalatine foramen was exposed. This usually required resection of the posterior half of the middle turbinate, a middle meatotomy and possibly a posterior ethmoidectomy, which allowed exposure right down to the orbital surface. Afterward, the posterior wall of the maxillary sinus was removed using a sphenoid punch. This procedure was usually not difficult in view of the curvature and thinning of the bone caused by the tumor. Thus, it is possible to remove virtually all of the posterior wall.
Endoscopy allowed clear identification between tumor and fat and therefore dissection of the internal maxillary artery was often possible. The latter then underwent diathermy, and the tumor was located medially. Even large tumors could then be pushed toward the cavum. The resection then continued at the level of the inferior orbital fissure toward the roof of the infratemporal fossa. After having checked for extension to the foramen ovale and rotundum, the resection was completed at the pterygoids. Extension to the base of the pterygoids, in the interpterygoid fossa or to the body of the sphenoid, may necessitate reaming of the pterygoid plates.
En bloc resection of the tumor was possible in 2 cases, even for 1 tumor with the greatest diameter of 7 cm. The tumor was pushed down into the pharynx, and the extraction was then orally performed.
A careful inspection of the operative field should be performed to detect any possible remnants. Additional coagulation was carried out as necessary. Packing of the nasal fossa was not performed in 16 cases, while packs were left in place for 2 days in 4 patients. Almost half of the patients in this series had extensive involvement of the infratemporal fossa.
Using the Radkowski classification, 9 patients had stage IIIA tumors; 7 patients, stage II (5, stage IIA; 1, stage IIB; and 1, stage IIC); and 4 patients, stage I (2, stage IA; 2, stage IB). In 2 patients, a remnant was detected at the time of the follow-up imaging at 6 months. In 1 case, the remnant was situated at the apex of the orbit (which was deliberately left in place at the time of surgery); this patient had stage IIIA tumor. In the other, the remnant was situated in the interpterygoid fossa; this patient had stage IIC tumor. These 2 patients had follow-up imaging at 18 and 20 months after surgery, which demonstrated that the remnants had not progressed.
The mean follow-up was 22 months, with a median of 15 months. The last follow-up scan (7 magnetic resonance imaging and 13 computed tomographic scans) was performed on average 16 months after surgery (median, 18 months). The average dimension of the resected tumors was 4.5 × 4 × 3 cm. The use of an exclusively endoscopic approach enabled a relatively quick surgical resection (mean duration of surgery was 135 minutes), that is, a shorter duration than would have been expected for the same tumors operated on by an open approach. Two patients in this series were operated on using computer-assisted navigation.
Blood loss was systematically monitored (measurement of the volume of blood aspirated). The mean volume of blood loss was 350 mL (range, 50-750 mL; median, 300 mL). In 4 cases, all of which were stage IIIA, there was greater than 500 mL of blood loss, which represents a significant difference with respect to stages I and II for which there were no cases of hemorrhage at this level (P<.02, χ2 test). Two of these 4 cases resulted from significant bleeding from the internal maxillary artery, while the other 2 resulted from bleeding from the body of the sphenoid. No patient required heterologous transfusion, and 1 patient who had congenital factor VII deficiency was operated on with replacement of this coagulation factor and without significant hemorrhage.
No other perioperative complications were noted. Furthermore, there was no postoperative hemorrhage. Erosion of the nares due to the traction of the instruments and possibly to passage of the motor (Figure 1) usually heals without leaving any trace. In 2 cases, there was anesthesia of the infraorbital nerve related to extensive adhesion of the tumor to the nerve in the inferior orbital fissure. In the present study, no additional treatment was necessary.
This series of 20 purely endoscopic resections of JNA demonstrates that a surgical endoscopic approach can be used for tumors of variable size. The treatment of JNA is associated with several of the following problems: the perioperative risk of hemorrhage; the difficulty of the resection; the morbidity associated with involvement of certain sites; and the risk of recurrence.
With respect to the risk of recurrence, staging is extremely important. The Radkowski classification allows prognostication based on the stage. Specifically, involvement of the base of the skull and in particular the roof of the infratemporal fossa, the cavernous sinus, and the body of the sphenoid are associated with an increased rate of relapse.7 Surgical endoscopy encounters the same difficulties as open surgery, in particular, at these sites.18,19 Nevertheless, in all of the published series in which an exclusively endoscopic approach has been used10- 20 and in our patients, no relapses were found in cases with stage I, IIA, and IIB tumors. The standard relapse rate for tumors staged I through IIB varies between 0% and 15%,2,3,6 which indicates that endoscopy is at least as effective as open surgery. In our series, of the 9 patients with Radkowski stage IIIA tumor, there was only 1 case in which a complete eradication was not possible (which was also the case for 1 patient with stage IIC tumor). Furthermore, these 2 remnants have not required any additional treatment (with a clinical follow-up of 36 and 30 months, respectively) and the success rate for stages IIC and IIIA would appear to be better than after open surgery: no true relapses occurred in the 10 cases in our series compared with the usual rate of 25% for stage IIC and 40% to 50% for stage III tumors.7,18 However, the relatively small number of patients with these large tumors that have been treated by an exclusively endoscopic approach means that it is not possible to be certain about this apparent reduction in the relapse rate. In addition, the radiological follow-up of our patients does not allow one to completely eliminate the possibility of a late relapse, even though this is a rare occurrence. This would require imaging for all patients for at least 2 years after surgery. The remnants following surgical excision of JNA may remain stable and may even spontaneously regress,4,5,7 and if they are asymptomatic, clinical and radiological surveillance is all that is required. If symptoms reoccur (epistaxis or nasal obstruction) or the lesions progress, appropriate treatment must be instigated, with further surgery or radiotherapy being the principal therapeutic modalities.
The resection of JNA is classically considered to be high-risk surgery with respect to hemorrhage and may require blood transfusion, which is not without morbidity. In addition, the perioperative bleeding can significantly hinder the excision. Consequently, most authors agree that preoperative embolization should be performed.23,24,26 Embolization was used in all the patients in the present study, apart from 1 patient who had a relapse with a small intrasphenoidal tumor following previous open surgery. The hemorrhagic risk is of course increased with congenital defects of hemostasis (eg, hemophilia27 or factor VII deficiency, as in 1 patient in our series).
The greater availability of procedures for autologous blood donation has allowed the use of heterologous transfusion to be avoided. However, collecting autologous units can delay surgery by several weeks and may be difficult to achieve if there is significant epistaxis. One patient in our series could only donate blood following the use of erythropoietin.
The risk of perioperative bleeding is greater at certain sites of dissection, for example, adjacent to the internal maxillary artery, the body of the sphenoid sinus, the roots of the pterygoids, the interpterygoid fossa as well as in the region of the cavernous sinus. Involvement of the cavernous sinus itself or extension toward the internal carotid artery through the foramen lacerum are associated with a risk of uncontrollable bleeding and probably represent the limits of endoscopy.7,18 To control bleeding in the region of the body of the sphenoid, it may be necessary to ream the pterygoid process of the sphenoid bone or even the sphenoid itself.
It is essential that the surgical team is adequately trained and has sufficient experience. One of the frequently mentioned problems with endoscopic surgery is the use of 1 hand to hold the endoscope. With the aid of videoendoscopy, it is possible for a trained assistant to position an aspiration tube at the correct site and thereby greatly facilitate the surgical procedure. The passage of 3 instruments via the nares does not usually lead to any problems even if slight damage occurs for a few days (Figure 1).
Invasion of certain structures leads to great difficulty with respect to access regardless of the type of approach used due to the narrowness of the access to the operative field, namely, the pterygoid recess of a sphenoid sinus with marked pneumatization or a very deep site in the operative field such as the clivus or the foramen lacerum. Similarly, the roots of the pterygoid and the interpterygoid fossa are also difficult to reach both by a transfacial or a lateral approach.
The use of a 30° telescope allowed this problem to be partly overcome owing to the additional angulation that could be achieved (Figure 2). The anterior limits of the field of exposure were constituted by the piriform aperture and the nasal septum. However, the latter can be easily pushed laterally, and the 30° angulation allows wide exposure of the retromaxillary region. Significant lateral extension into the infratemporal fossa, especially toward the area of the cheek, is probably not accessible by endoscopy alone. Even using more angulated equipment (45° or 70°), it is unlikely that the instruments would be able to satisfactorily reach such lateral invasion. In the roof of the infratemporal fossa, the possible interdigitating penetration by the tumor would make resection very difficult regardless of the surgical technique used. Although endoscopy undoubtedly provides better visualization in this part of the operative field, penetration of the temporal fossa itself requires an approach in which the surgeon has both hands free. The angle of vision provided by endoscopy is greater than that with open surgery because it enables very close inspection of the tumor in all directions. One zone, however, remains very difficult to expose initially, namely, the posterior part of the tumor. Consequently, we propose that it is best to start with mobilization of the anterior and lateral parts of the tumor to access its posterior connections as laterally as possible.
The progressive detachment and traction of the tumor allow progression to the posterior regions, namely, the body of the sphenoid, the roots of the pterygoids, and the interpterygoid fossa. Even in these sites, the view with endoscopy is better than that with open surgery. Furthermore, careful reappraisal is easier with endoscopy in view of the better visual field and the ability to aspirate selectively.
Combining endoscopy with open surgery is another interesting possibility.7,20 This would have the advantages of both techniques (the quality of the field of exposure and the availability of both hands) but would have the disadvantage of the morbidity associated with open surgery (possible scarring, duration of the surgery, and risk of increased hemorrhage). The choice between an exclusively endoscopic approach and surgery facilitated by endoscopy depends on the localization of the tumor as well as the experience of the surgeon. Figure 3, Figure 4, Figure 5, and Figure 6 illustrate the resection of JNA by a endoscopic approach.
An exclusively endoscopic surgical approach was associated with a shorter duration of hospitalization (most of our patients were able to be discharged within 48 hours after surgery). In addition, this avoids all of the complications specifically related to different approaches with open surgery (epiphora, dysesthesia, trismus, pain, and possible effects on facial growth). These advantages, combined with the reduction in the duration of surgery (related to the absence of a surgical incision) and especially the expected reduction in perioperative bleeding, do not of themselves constitute adequate reasons for considering a purely endoscopic approach as being the first choice for the resection of JNA. Nevertheless, the published results10- 20 have demonstrated an identical efficacy to that of open surgery, at least for stages I and II and probably for some stage III cases, without intracranial involvement.
The development of computer-assisted endoscopic surgery (2 cases in the present series) should allow improved landmarking of the tumor and consequently make endoscopy easier to perform.28 The use of a potassium-titanyl-phosphate laser15 may possibly help achieve good hemostatic control during the operation. Further development of the endoscopic approach will also lead to a greater number of surgeons becoming familiar with this procedure for treating JNA and thereby allow a progressive extension of the indications for an exclusively endoscopic management of small, medium, and even large tumors.
The resection of JNA by a purely endoscopic approach is a reliable technique, and the results, in terms of excision, are certainly as good as with open surgery. Even Radkowski stage IIC and IIIA tumors can be removed using this technique, although the patient must be informed that it may be necessary to resort to an open procedure. For lower-stage tumors, the efficacy of endoscopy has been established. Detailed analysis of high-quality imaging as well as embolization are essential preoperative requirements. Similarly, it is essential that the entire surgical team is highly experienced with endoscopic techniques. The presence of 2 experienced surgeons allows the use of 3 hands in the operative field, thus considerably facilitating operative gestures. Endoscopic surgery allows quicker resection with less hemorrhage and is very effective. The areas where excision is difficult are the same and may lead to the risk of residual tumor remnants, which can be simply observed. The recurrence of symptoms or progression of lesions may require further treatment.
Accepted for publication January 18, 2002.
This study was presented at the 16th annual meeting of the American Society of Pediatric Otolaryngology, Scottsdale, Ariz, May 11, 2001.
Corresponding author and reprints: Gilles Roger, MD, Pediatric Ear, Nose, and Throat Department, Armand-Trousseau Children's Hospital, 26 Avenue Arnold Netter, 75571 Paris CEDEX 12, France (e-mail: firstname.lastname@example.org).