CLINICAL STUDIES

MANAGEMENT OF INVASIVE JUVENILE

NASOPHARYNGEAL ANGIOFIBROMAS:
THE ROLE OF A MULTIMODALITY APPROACH
Pierre-Hugues Roche, M.D.
Department of Neurosurgery,
Sainte Marguerite University Hospital,
Marseille, France
Jerôme Paris, M.D.
Federation of Ear, Nose,
and Throat Surgery,
La Timone University Hospital,
Marseille, France
Jean Régis, M.D., Ph.D.
Department of Stereotactic and
Functional Neurosurgery,
La Timone University Hospital,
Marseille, France
Guy Moulin, M.D.
Department of Radiology,
La Timone University Hospital,
Marseille, France
Michel Zanaret, M.D.
Federation of Ear, Nose,
and Throat Surgery,
La Timone University Hospital,
Marseille, France
Jean-Marc Thomassin, M.D.
Federation of Ear, Nose,
and Throat Surgery,
La Timone University Hospital,
Marseille, France
William Pellet, M.D.
Neurosurgery 61:768–777, 2007
Department of Neurosurgery,
Sainte Marguerite University Hospital,
Marseille, France
OBJECTIVE: Juvenile nasopharyngeal angiofibromas involving the cranial base and
intracranial compartment are challenging tumors. We reviewed our experience of these
tumors and analyzed the efficacy of a multimodality management.
METHODS: Between 1981 and 2000, 15 extensive juvenile nasopharyngeal angiofibro-
mas (Fisch Grade III or IV) were treated at our institution. The mean age of the patients was
14.5 years, and the mean interval between the first symptom and diagnosis was 12.9
months. Initial management included preoperative embolization of the external carotid
artery feeders, followed by tumor removal. A maxillofacial procedure was performed in
eight cases, a combination of maxillofacial and neurosurgical approach was performed
in four cases, and a neurosurgical cranial base approach was performed in three cases.
RESULTS: Total removal after the initial procedure was obtained in eight patients.
Subtotal removal justified additional surgery in one case, gamma knife radiosurgery in
two cases, and fractionated irradiation in four cases. True recurrences were observed
in four cases at a mean interval of 37 months (range, 24–46 mo) and required tailored
multimodality management. No cases of perioperative death were observed. One
patient underwent hemiparesis after embolization in the early period of our experi-
ence. Permanent facial numbness was reported in four cases, moderate cosmetic prob-
lems were reported in three cases, and hyposmia was reported in three cases. Except
for one patient who was lost to follow-up at 18 months, 12 patients were free of dis-
ease and two patients were free of tumor progression. All patients had normal or near-
normal daily life at the last check-up, with a median follow-up period of 108 months
(mean, 117 mo; range, 91–252 mo).
CONCLUSION: Extensive juvenile nasopharyngeal angiofibromas are efficiently man-
aged with a multimodal protocol in which preoperative embolization is followed by
optimal surgical removal using various transcranial or transfacial approaches. Adjunctive
gamma knife radiosurgery is a valuable option for intracavernous residual tumor. Our
protocol offers long-term cure with acceptable morbidity.
KEY WORDS: Angiofibroma, Cranial base surgery, Neurosurgery, Radiosurgery, Skull base surgery
DOI: 10.1227/01.NEU.0000280077.89703.AD www.neurosurgery-online.com

Reprint requests:
Pierre-Hugues Roche, M.D.,
Service de Neurochirurgie,
Hôpital Sainte-Marguerite,
Assistance Publique des
Hôpitaux de Marseille,
Boulevard de Sainte-Marguerite,
Marseille, France.
Email: prochemail.ap-hm.fr
Received, December 26, 2006.
Accepted, May 1, 2007.
J uvenile nasopharyngeal angiofibromas
(JNA) are benign tumors affecting young
adolescent males and originating from the
posterolateral wall of the nasal cavity. In the
situation of early diagnosis, the tumor exten-
sion is limited and radical cure can be expected
using noninvasive surgical approaches (23).
However, in 20 to 40% of the cases, the tumor
is diagnosed at the time of cranial base exten-
sion with a potential involvement of intracra-
nial structures. In these cases, management is
challenging because of the relationship of the
tumor to the critical neurovascular structures
and to extensive vascular recruitment. There-
fore, a high neurological morbidity and a high
rate of incomplete resection may be expected.
Most neurosurgeons have little experience
with this disease. In this retrospective study,
we sought to identify the patterns of intracra-
nial extension of JNAs. We attempted to clarify
the respective indications of cranial base tech-
niques and the potential role of adjunctive
noninvasive procedures with special interest
to radiosurgery.

768 | VOLUME 61 | NUMBER 4 | OCTOBER 2007 www.neurosurgery-online.com

 INVASIVE JUVENILE NASOPHARYNGEAL ANGIOFIBROMAS

patients harbored focal tumors (Class I or II). Fourteen of these patients

TABLE 1. Fisch system for the grading of juvenile nasopharyngeal
angiofibromas
Type I Tumor limited to the nasopharynx and nasal cavity.
Bone destruction negligible or limited to the spheno-
palatine foramen.
Type II Tumor invading the pterygopalatine fossa or the maxillary,
ethmoid, or sphenoid sinus with bone destruction.
Type IIIa Tumor invading the infratemporal fossa or orbital region
without intracranial involvement.
Type IIIb Tumor invading the infratemporal fossa or orbit with
intracranial extradural (parasellar) involvement.
Type IVa Intracranial intradural tumor without infiltration of the
cavernous sinus, pituitary fossa, or optic chiasm.
Type IVb Intracranial intradural tumor with infiltration of the
cavernous sinus, pituitary fossa, or optic chiasm.
PATIENTS AND METHODS
Patient Population
This series represents our surgical experience with 15 patients
treated at the University Hospital of Marseille between 1981 and 2000.
This subgroup of patients was extracted from a larger population of
35 consecutive patients with JNAs who had been treated during the
same period. The present review encompasses 15 male patients harbor-
ing an extensive tumor mass involving the cranial base or the intracra-
nial compartment that could be classified as Class III or IV tumors
based on the Fisch (13) classification (Table 1), whereas the 20 other
were managed from the onset, and one patient came from another cen-
ter after tumor recurrence. Operative records, neuroimaging results,
and clinical histories, including postoperative follow-up data, were
reviewed. The mean age of patients at the time of diagnosis was
14.5 years (range, 11–20 yr), and the mean interval between the first
symptom and diagnosis was 12.9 months (range, 1–26 mo). The initial
evaluation of the patient included a general history, physical examina-
tion, and endoscope examination of the nasal and pharyngeal cavities.
The first symptoms were epistaxis (12 patients) and nasal obstruction
(13 patients), followed by proptosis, pharyngeal obstruction, hearing
loss, and trigeminal symptoms. The details of the clinical presentation
before our management are shown in Table 2.
Radiological Evaluation
Contrast-enhanced computed tomographic (CT) scans, magnetic res-
onance imaging (MRI) scans, or both were studied before surgery for
all patients. The tumor mass was characterized by a well-delineated
and high-density lesion that displayed a homogeneous enhancement.
The bony deformation and the erosion of the cranial base were shown.
MRI scans were particularly useful in analyzing the intracranial and
intraorbital extension. For each patient, tumor extension (Fisch class) is
given in Table 2. In addition, detailed radiological extensions to the dif-
ferent compartments of the cranial base are analyzed and displayed in
Table 3. In all patients, a carotid arteriogram was performed, with selec-
tive catheterization of external and internal carotid arteries on both
sides. Data available in 14 cases indicated that main feeders came from
the ipsilateral internal maxillary artery in all patients with a predomi-
nant role played by the sphenopalatine artery. Tumor feeders came
from the ascending pharyngeal artery in four cases, the ipsilateral inter-
nal carotid artery in five cases (mainly intracavernous branches), the
contralateral external carotid artery in five cases, and the contralateral

TABLE 2. Summary of clinical findings of juvenile nasopharyngeal angiofibromas in 15 patientsa

Patient Age at Initial Other Physical Interval between symp- Fisch
no. diagnosis (yr) symptom symptoms findings tom and diagnosis (mo) stage
1 11 E FD 30 IVa
2 12 E NO 18 IVa
3 15 E NO FD, P hypoacousia 14 IVa
4 11 NO 2 IIIa
5 13 E NO, headaches P, FD 1 IIIa
6 18 E NO 5 IIIa
7 17 E NO, HS 12 IIIa
8 12 E Headaches FD 9 IIIa
9 17 E NO 24 IVa
10 15 E FD, HS Pharyngeal mass 12 IVa
11 14 E NO 3 IIIa
12 13–20b E NO Hypoacousia 26 IIIa
13 16 E NO, headaches P, visual deficit, FD 19 IVb
14 13 NO HS 12 IIIa
15 14 E NO FD, facial numbness 6 IVa

a
E, epistaxis; FD, facial deformity; HS, hyponasal speech; NO, nasal obstruction; P, proptosis.
b
Thirteen-year-old boy at first management in another center and 20 years old at the time of our management.

NEUROSURGERY VOLUME 61 | NUMBER 4 | OCTOBER 2007 | 769

ROCHE ET AL.

TABLE 3. Tumor extensions shown by preoperative radiological workup

Radiological extension of the 15 juvenile nasopharyngeal angiofibromas
Orbit
Cranial base structures Intracranial structures
Pterygoid plate 15 Extradural temporal fossa 5 6
Pterygomaxillary fissure 15 Intradural temporal fossa 1
Infratemporal fossa 8 Superior orbital fissure 2
Posterior parapharyngeal space 6 Cavernous sinus 4
Sphenoid sinus 12
Inferior orbital fissure 10

internal carotid artery in two

cases. Postoperative MRI or CT
scans were obtained within 2
months of surgery to determine
the extent of resection in each
patient and in the long-term fol-
low-up period to assess a poten-
tial recurrence.

Extent of Resection
The determination of extent of
resection was based on the intra-
operative assessment, postopera-
tive findings of the fiberoptic
examination of the nasal and pha-
ryngeal cavities, and postopera-
tive images (Fig. 1). Gross total
removal was defined as a patient
displaying no intraoperative evi-
dence of residual tumor, normal
fiberoptic examination, and no
evidence of contrast-enhancing
tumor on postoperative imaging.
The cases of residual tumor were
considered as partial removal.
Tumor recurrence was defined as
any newly identified enhance-
ment after gross total removal.

Preoperative Embolization FIGURE 1. Diagram illustrating the treatment protocol used in our population. JNA, juvenile nasopharyngeal
angiofibroma; MF, maxillofacial; PLN, paralateronasal; MD, midfacial degloving; mPSIT, modified preauricular sub-
Several agents were used to
and infratemporal approach; ETP, epidural temporopolar approach; GKR, gamma knife radiosurgery; FUp, follow-
selectively occlude the feeders that
up; RT, radiotherapy
came from the external carotid
artery from one or both sides:
Histoacryl glue (B-Braun, Melsungen, Germany), resorbable Gelfoam tumor resection, taking into consideration the high vascularity, the
(Upjohn Co., Kalamazoo, MI), or a combination of Gelfoam with flower locoregional extension, and the fibrous consistency of the tumor. We
microcoils (Guerbet Biomedical, Louvres, France). The procedure was always attempted to achieve an en bloc extracapsular resection using
delivered within the 72 hours before surgical removal (Fig. 2). extensive surgical corridors. A meticulous hemostasis was permanently
performed to reduce blood loss. The attempt of intracavernous resec-
tion of tumor fragments was never undertaken, avoiding the mobiliza-
Approach Selection and Operative
tion of critical neurovascular structures.
Technique in First Intention Ear, nose, and throat approaches were the transfacial approach with
The surgical approach was predicated on the location of the epicen- a paralateronasal skin incision, the transmaxillary transantral approach,
ter of the tumor, direction of tumor extension, and tumor size (Table 4). and the transmaxillary approach via midfacial degloving. These
Regardless of the approach, we used the same general principles of approaches have been extensively detailed in other articles (19).

770 | VOLUME 61 | NUMBER 4 | OCTOBER 2007 www.neurosurgery-online.com

 INVASIVE JUVENILE NASOPHARYNGEAL ANGIOFIBROMAS

Neurosurgical approaches consisted of various temporobasal cran-

A B
C D
FIGURE 2. Radiological data from Patient 9. A, preoperative angiogram
showing the numerous feeders coming from the branches of the internal
maxillary artery. B, coronal CT scan indicating the extension of the tumor
mass to the paranasal sinuses, inferior orbital fissure, and temporal fossa
on the right side. C, after selective embolization of the internal maxillary
artery by coils, feeders are still coming from the intracavernous branches
of the internal carotid artery. D, postoperative CT scan showing the extent
of resection and lack of residual tumor.
iotomies. We initially adopted a modified preauricular sub- and
infratemporal approach, derived from Sekhar et al.’s (26) technique.
The approach included a preauricular skin incision followed by the
mobilization of the temporal muscle after zygomatic deposition. The
temporomandibular joint was translocated anteroinferiorly and a tem-
poral bone flap was made, including the optimal resection of the floor
of the middle fossa as far as the pterygoid plate. The dissection of the
upper neck was then undertaken, which exposed the division of the
facial nerve inside the parotid gland and the internal carotid artery
under the petrous bone. After optimal resection of the tumor, including
the components that involved the infratemporal fossa and the parapha-
ryngeal space, meticulous reconstruction and closure were performed.
More recently we adopted a less invasive technique, the epidural tem-
poropolar approach, derived from the description of Hakuba et al. (15).
The patient was positioned supine with the head rotated 45 degrees to
the opposite side. The skin incision was started 1 cm inferior to the
zygoma and 1 cm anterior to the tragus, prolonged upward in a ques-
tion mark fashion, ending behind the hairline in the frontal area. The
pericranial flap was elevated and turned downward and forward. The
supraorbital nerve was freed from the supraorbital foramen and
reflected within the flap. The temporalis muscle and fascia were dis-
sected subperiosteally. A temporopterional bone flap was made and
included an orbitozygomatic deposit in a “one-piece” manner (Fig. 3).
The frontotemporal dura, periorbit, and superior orbital fissure were
widely exposed and could be opened in the case of intradural involve-
ment. The floor of the temporal fossa was drilled extensively to expose
the foramen rotundum and ovale. The upper part of the infratemporal
fossa, pterygopalatine fossa, sphenoid sinus, and posterior wall of the
maxillary sinus could also be exposed, depending on tumor extension.

TABLE 4. Synopsis of the management and follow-up of patientsa

Patient First Extent of Adjunctive Delay of recurrence/ Length of follow- MRI at
no. treatment resection treatment additional treatment up (yr after surgery) last follow-up
1 PLN STR RT —/— 1.5 (LFU)
2 PLN and mPSIT STR GKR —/— 4 DRT
3 mPSIT STR RT —/— 20 NRT
4 Transmaxillary GTR — 46 mo/MD and GKR 7 DRT
5 PLN GTR — 44 mo/PLN and mPSIT 23 NRT
6 MD GTR — —/— 8 NRT
7 PLN GTR — 30 mo/PLN 11 NRT
8 PLN STR RT —/— 8 NRT
9 MD and mPSIT GTR — —/— 5 NRT
10 mPSIT GTR — 24 mo/PLN 4 NRT
11 PLN GTR — —/— 11 NRT
12 PLN GTR — —/— 12 NRT
13 PLN and mPSIT STR RT —/— 12 NRT
14 mPSIT STR MD —/— 12 NRT
15 PLN and ETP STR GKR —/— 5 NRT

a
MRI, magnetic resonance imaging; PLN, paralateronasal approach; STR, subtotal removal; RT, radiation therapy; LFU, lost to follow-up; mPSIT, modified preauricular sub-
and infratemporal approach; GKR, gamma knife radiosurgery; DRT, decreased residual tumor; NRT, no residual tumor; GTR, gross total removal; MD, midfacial degloving;
ETP, epidural temporopolar approach.

NEUROSURGERY VOLUME 61 | NUMBER 4 | OCTOBER 2007 | 771

ROCHE ET AL.
FIGURE 3. Interoperative photograph of Patient 15
showing the epidural temporopolar approach that was
conducted simultaneously to the paralateronasal proce-
dure. One-piece orbitozygomatic pterional bone flap is
achieved (inset), allowing an access to the infratempo-
ral fossa. The upper and lateral pole of the tumor is in
the inferior orbital fissure (white arrow). GSW, great
sphenoid wing; OF, orbital fascia.
Case Selection for External Beam Radiation
Therapy and Gamma Knife Radiosurgery
From 1981 to 1992, before the development of radiosurgery in our
institution, external beam radiation therapy was delivered in case of
residual tumors unable to be extirpated. Delivered doses ranged from
25 to 50 Gy (mean dose, 40 Gy; median dose, 37.5 Gy), with a 2 Gy per
fraction protocol. From 1992 to 2000, gamma knife radiosurgery
(GKR) was administered after incomplete resection when the residual
tumor mass did not exceed a 3 cm diameter and the optic pathway
was not in the close proximity of the lesion (Fig. 4). Tumors were
treated using a multi-isocentric protocol. The dose that was deliv-
ered at the tumor margin ranged from 14 to 16 Gy, corresponding to
the 50% isodose (Table 5).
Follow-up Protocol
Insofar as it was possible, the patients underwent a sequential fol-
low-up evaluation by our team, including iterative clinical and fiberop-
tic controls and CT or MRI scans. For the purpose of this study, we
were able to collect recent information for all patients except one, with
particular attention to their level of autonomy and quality of life.
TABLE 5. Summary of juvenile nasopharyngeal angiofibroma cases who underwent a gamma knife radiosurgical procedurea
Patient
Indication of GKR
no.
2 Residual tumor after initial surgery MCC
4 Residual after second surgery MCC, inferior orbital fissure
for recurrence
15 Residual tumor after initial surgery Whole cavernous sinus
a
GKR, gamma knife radiosurgery; MCC, medial cavernous compartment; NC, no complication.
772 | VOLUME 61 | NUMBER 4 | OCTOBER 2007
A B
C D
FIGURE 4. MRI scans of Patient 15. A, the sphenoid sinus, infratempo-
ral fossa, and intracranial parasellar compartment are involved by the
tumor. B, preoperative axial view, the whole ethmoid is involved. C, early
postoperative axial view indicating a residual tumor mass (arrow) in the
right cavernous sinus. D, MRI scan obtained 5 years after GKR showing
that there is no evidence of residual tumor mass.
RESULTS
One patient (Patient 1) was lost to follow-up at 28 months
after surgery. For the 14 remaining patients, the median clinical
and radiological follow-up period was 108 months (mean, 117
mo; range, 91–252 mo) after the first treatment in our institution.
Embolization
Thirteen patients (all patients except Patients 5 and 8) under-
went preoperative endovascular tumor embolization. The fol-
lowing arteries were embolized: internal maxillary artery on
one side in 10 patients, internal maxillary artery on both sides
in five patients, ascending pharyngeal artery in nine patients,
facial artery in two patients, and accessory meningeal artery in
four patients. This management allowed a significant reduction
of the blood loss during surgery. Permanent hemiparesis was
Location of
Dosimetry Outcome
remnant tumor
16 Gy at the 50% peripheral Decrease at 4 yr
isodose (isodose 50) NC
14 Gy isodose 50–11 Decrease at 3 yr
isocenters NC
14 Gy isodose 50–36 No residual at 5 yr
isocenters NC
www.neurosurgery-online.com

observed immediately after embolization in Patient 1 and was
caused by the migration of the embole product to the middle
cerebral artery during the early stage of our experience. In
another patient (Patient 13), the embolization may have con-
tributed to the formation of a limited scalp necrosis that
occurred after surgery; however, the long-term cosmetic result
was acceptable and did not require any additional procedures.
Tumor Removal
The first-intention surgical strategy was distributed as fol-
lows (Fig. 1; Table 4): four patients underwent a combined max-
illofacial and neurosurgical approaches in a one- or two-stage
procedure. Eight patients were managed by a single maxillofa-
cial approach, and a single neurosurgical procedure was per-
formed in the three remaining patients.
Gross total removal of the tumor mass was obtained in eight
of the 15 patients (53.3%; confidence interval, 26.6–78.7).
Subtotal removal was observed in seven of the 15 (46.7%)
patients after the first procedure and mainly involved patients
with Fisch Class IV tumors. In six patients, the remnant tumor
was not eligible for a more aggressive attempt of resection after
careful consideration of the risk of additional cranial nerve
injury and potential vascular damage. Fractionated radiation
therapy was performed in four patients, and GKR was per-
formed in two patients (Table 5). In the last patient, a new sur-
gical step (midfacial degloving) was performed without
adjunctive irradiation. On average, the estimated blood loss
during surgery was 1037.5 mL (standard deviation, 1091 mL).
Recurrence
True recurrence of the disease was observed in four of the
eight patients (50%) in whom gross total removal was
achieved. The mean interval between the first treatment and
recurrence was 37 months (range, 2–4 yr) after surgery. In all
cases, new symptoms (epistaxis and nasal obstruction) were
contemporary with this event. On the basis of the endoscopic
evaluation and postoperative CT and MRI scans, the location
of recurrence was focal in two patients (choana and ptery-
gopalatine fossa) and more extensive (middle cranial base
and infratemporal fossa) in the other two patients. Treatment
of the recurrence consisted in additional surgical procedure in
all patients, followed by GKR in one patient. In this small
group of patients, it was not possible to identify predictive
factors that may have influenced the recurrence, particularly
in respect to the age of the patient, tumor extension, and Fisch
class at the time of management. Among the four recurrent
cases, three patients had undergone a single maxillofacial
approach as the initial procedure.
Complications
Eight patients from this series displayed at least one perma-
nent problem resulting from the surgical treatment of their dis-
ease. Trigeminal hypesthesia, especially in the area of the max-
illary branch, was observed in four patients, three of whom
had undergone an extensive combined maxillofacial and neu-
rosurgical approach, including a modified preauricular sub-
NEUROSURGERY
INVASIVE JUVENILE NASOPHARYNGEAL ANGIOFIBROMAS
FIGURE 5. Histological features of JNA showing that
the tumor was composed of fibrocollagenous stroma
containing vascular spaces (hematoxylin and eosin;
original magnification, ⫻40).
and infratemporal approach. Cosmetic problems (unesthetic
skin scar) were encountered in two of the 10 patients who
underwent operation via a paralateronasal approach. One of
them displayed a permanent enophthalmia after removal of
the intraorbital tumor fragment. Another patient had perma-
nent rhinitis with hyposmia and a bird-like nose deformation
responsible for psychoaffective disturbances after a midfacial
degloving. Permanent hyposmia was experienced by two other
patients after a maxillofacial procedure. Chewing difficulties
were reported by one patient who had undergone a modified
preauricular sub- and infratemporal approach with resection of
the mandibular condyle.
No deaths were reported during the exhaustive follow-up
period for 14 patients, and all patients are now living with
what they estimate as a satisfactory level of quality of life,
although a dedicated questionnaire has not been filled out.
DISCUSSION
General Considerations
The term “juvenile nasopharyngeal fibroma” was introduced
by Chaveau (4) in 1906 to describe a nodular and dark tumor
of the choana. JNA is a nonencapsulated, highly vascular lesion
originating from the mesenchymal tissue. Compression of sur-
rounding normal tissue produces a pseudocapsule of fibrous
tissue. The tumor is histologically benign and composed of a
connective tissue stroma with a matrix of a dilated vascular
structure (Fig. 5). The blood vessels usually lack a complete
muscular layer, which may account for the lesion’s well-known
hemorrhagic tendencies (28). JNAs occur almost exclusively in
adolescent males and account for only 0.5% of all neoplasms of
the head and neck, but they remain the most common benign
tumor of the nasopharynx. The main symptoms are nasal
obstruction in 90% of the patients and spontaneous epistaxis
(atraumatic recurrent epistaxis) in 60% of the patients. When
VOLUME 61 | NUMBER 4 | OCTOBER 2007 | 773
ROCHE ET AL.
locoregional extension to the cranial base is observed, patients
may display facial deformation, proptosis, and cranial nerve
deficits. Jafek et al. (18) reported proptosis in 20% and cranial
neuropathy in 33% of 15 patients with JNAs involving the mid-
dle cranial fossa.
Natural History and Tumor Extension
JNAs are thought to arise at the site of the junction of the
membranous viscerocranium (palatine, vomer, pterygoid) and
the cartilaginous neurocranium (sphenoid). Anatomically, the
site of development is the posterolateral wall of the nasal cav-
ity, close to the sphenopalatine foramen. JNAs tend to extend
along natural foramina and fissures, spreading toward the
paranasal sinus (ethmoid, sphenoid, maxillary) and through
the pterygopalatine fossa laterally. This growth in soft tissue
spaces is relatively unrestricted, limited only by bone that is
usually not invaded but rather eroded by pressure. This phe-
nomenon is well illustrated by CT images of “maxillary antrum
bowing,” usually observed in cases involving the pterygomax-
illary fissure. If there is further progression, regional involve-
ment laterally joins the infratemporal fossa, backward toward
the parapharyngeal space. Upward, the extension to the orbit is
done through the inferior orbital fissure, but the orbital fascia
is not involved in most cases.
Intracranial extension was reported in 10 to 20% of all cases
before the era of modern neuroimaging. With the use of high-
resolution CT and MRI scans, this incidence increases to 36%.
Several routes of intracranial involvement are described. The
tumor may involve the parasellar compartment through the
superior orbital fissure. In many cases, the orbital involvement
lies outside the rectus muscle cone and may extend to the mid-
dle fossa without involving the dura. As outlined in previous
studies (17, 21), deep extension into the cancellous bone at the
base of the pterygoid process is often associated with expan-
sion and invasion of the greater sphenoid wing leading to a
direct invasion of the middle fossa. Several natural channels,
namely, the vidian canal, foramen rotundum, and foramen
ovale, perforate the middle fossa and represent the natural
pathway of intracranial invasion. It is of note that intradural
invasion is rare and brain invasion has never been observed. In
a review of 84 cases of intracranial JNA by Jones et al. (20), only
one case of dural penetration was reported. In our own expe-
rience, we observed a unique case of intradural extension
among 15 cases of invasive JNAs, and histological examination
did not show differences in the subgroup of focal JNAs com-
pared with the subgroup of invasive JNAs (data not shown).
Primary Treatment of Invasive JNAs
The rate of radical cure of focal JNAs is approximately 100%.
This result can be achieved with reduced morbidity and using
ear, nose, and throat approaches, such as lateral rhinotomy, the
transpalatal approach, or more recently, endoscopic excision.
The purpose of this article is to show how a multimodality
management is able to achieve satisfactory results in invasive
JNAs. This study outlines the importance of a thorough preop-
erative workup of the disease to plan the different steps of the
774 | VOLUME 61 | NUMBER 4 | OCTOBER 2007
treatment. Clinical examination is followed by an endonasal
fiberoptic endoscopy. Taken together, CT scans with bone win-
dow and MRI scans evaluate the infratemporal, paranasal,
intraorbital, and intracranial extensions of the disease. A bilat-
eral carotid digital substraction angiography with selective
catheterization of the external and internal carotid arteries is
required to describe the tumor blood supply. Preoperative
embolization of the external carotid artery is the first step of the
treatment. The occlusion of the tumor’s feeders provokes the
progressive development of thrombosis in the small distal ves-
sels of the tumor, allowing for tumor shrinkage and an easier
en bloc removal. McCombe et al. (22) stated that the quality of
preoperative embolization was essential to prevent recurrence.
In the present series, embolization did not involve the feeders
coming from the internal carotid artery because of high risk of
stroke. This explained the remnant tumor vascularity and
residual bleeding during the surgical step. Cranial nerve
palsies, cutaneous and subcutaneous necrosis, and local pain
lasting a few days are sometimes reported after embolization
(29). In a study by Zhang et al. (29) no permanent sequela was
noted after 20 embolized JNA cases, but one patient experi-
enced a partial visual loss that was recovered within 7 days. In
the early stage of our experience, one of our patients experi-
enced a hemiparesis as the result of agent migration in the
internal carotid artery.
The essential step of the management is surgical removal. It
has clearly been shown that the extent of initial surgery is cru-
cial to prevent recurrence (21). However, this surgery is chal-
lenging because of the depth to reach, the pattern of vascular-
ity, and the multiplicity of extensions to the skull base and
intracranial structures. Because almost all tumors remain
extradural, single-stage maxillofacial extracranial approaches
have been proposed for even very large tumors (20). However,
the depth of the operative field, difficulties controlling the
bleeding from the cavernous space, and potential disturbance
of bony facial development are well-known limitations of this
strategy. In our personal experience, these approaches provided
a satisfactory rate of radical resection with acceptable compli-
cations in selected cases.
Wide cranial base and intracranial extension of JNAs may
require extensive cranial base procedures. The infratemporal
approaches pioneered by Andrews et al. (1), Fisch (13), Fisch
and Mattox (14), and Zhang et al. (29) offer a direct and wide
exposure of the infratemporal fossa and parapharyngeal space
together with the middle fossa and cavernous sinus with a
shortened working distance. Such procedures resulted in an
80% rate of radical cure and a 6% incidence of recurrence.
However, resection of the glenoid fossa is responsible for mal-
occlusion and long-term mandibular dysfunction. The division
of the mandibularly and possibly of the maxillary branch of the
trigeminal nerve entails facial anesthesia and sometimes
dolorosa anesthesia. Slight temporal depression and conductive
hearing loss are also two notable permanent symptoms in a
Type C approach. The same principle, efficacy, and drawbacks
may be attributed to the preauricular sub- and infratemporal
approach described by Sekhar et al. (26). An alternative to these
www.neurosurgery-online.com

invasive procedures is to perform a fronto-orbitozygomatic
craniotomy with an epidural temporopolar approach, as illus-
trated in one of our cases. This allows the control of large
intracranial and cranial base involvement but offers a restricted
approach to the intrapetrous and parapharyngeal space. Di
Rienzo et al. (8) recently proposed a modification of this latter
approach. In their experience, the so-called open-mouth fronto-
orbitozygomatic approach allowed the joining of the sphenoid
sinus, pterygoid plate, and pterygopalatine fossa using a more
downward retraction of the coronoid process. They published
three cases of high-grade JNAs that were totally removed using
this technique.
As shown by our experience and depending on the tumor
extension, it may be necessary to associate both maxillofacial
and cranial base neurosurgical approaches simultaneously or in
a two-stage strategy. These combined procedures offer multiple
operative corridors, better vascular control, and a better chance
to achieve radical resection. However, such procedures add to
the risk and morbidity of both approaches.
Management of Residual Tumor after Surgery
After the initial surgery, a thorough clinical and radiological
workup should be conducted because incomplete cure can lead
to potential regrowth of the remnant tumor. In published
works, the treatment of residual tumor is debated, most likely
because the biological behavior of such remnants is difficult to
predict. In their series, Herman et al. (16) observed residual
tumors in nine out of 44 patients, and a prolonged follow-up
period showed a stable or decreasing mass in all of these. In the
same manner, English et al. (11) described four cases of incom-
plete extracranial resection with angiographically proven
regression of residual tumor. Dohar and Duvall (9) also
reported the same experience. Zhang et al. (29) deliberately left
intracavernous remnant tumors in four out of 20 patients
undergoing operation and did not observe any tumor growth
on serial imaging during a 46- to 53-month follow-up period.
This favorable behavior is not always observed, particularly in
young patients. We personally consider that, in the case of
symptomatic tumors occurring in young adolescents, the JNA
is in an active state of growth and, thereby, justifies active treat-
ment of the residual fragment. Therefore, we systematically
undertook adjunctive treatment. For extracavernous tumors
able to be extirpated, we preferred additional surgery. The
problem is more debatable for intracavernous remnant tumors
because they are usually poorly symptomatic and difficult to
remove safely. Close et al. (5) reported six cases of JNAs pre-
senting with cavernous sinus involvement for which the
patients underwent radical surgery. Total removal was
obtained in all patients using various approaches. However,
the authors reported a postoperative permanent trigeminal
hypesthesia in three patients and oculomotor palsy in one. This
experience was not reported elsewhere in the literature and
was conducted before the advent of magnetic resonance neu-
roimaging, giving an imprecise view on the actual intracav-
ernous location of the tumor. As advocated by Fisch (13) and
Jafek et al. (18), it may be safer to adopt nonsurgical treatment
NEUROSURGERY
INVASIVE JUVENILE NASOPHARYNGEAL ANGIOFIBROMAS
for intracavernous residual tumors and choose an adjunctive
radiation therapy technique. Efficacy of radiation therapy in
JNAs was initially documented in studies in which it had been
used in first intention (2, 6, 10). On average, an 80% long-term
tumor control rate was obtained using doses between 30 and 35
Gy. However, in the series of 55 cases reported by Cummings
et al. (6), one patient developed a thyroid carcinoma and one
patient developed a facial basal cell carcinoma. More recently,
Reddy et al. (25) reported three cases of cataracts and one case
of basal cell carcinoma of the skin in a series of 15 patients.
Because of these potential complications and the risk of radia-
tion-induced neuropathy, hypophyseal insufficiency, or inter-
ference with development on facial skeleton, radiation therapy
should be reserved for unresectable remnant disease in this
age group. In our population, adjunctive radiation therapy was
motivated by the extent of the residual tumor despite an opti-
mal attempt of resection and considering the growing potential
of these lesions. This treatment was given at the initial step of
our practice and, although we did not have to experience any
radio-induced complication, the use of stereotactic radio-
surgery was later preferred in our group. Well-vascularized
and delineated residual tumors seem to be a suitable target for
the gamma knife, particularly in cases of an intracavernous
location (Fig. 4). Recently, Dare et al. (7) reported their experi-
ence of two patients treated with GKR using 20 Gy at the tumor
margin in both cases. Both tumors were small residual cranial
base JNAs treated in the months after surgery. The follow-up
periods were 2 and 3 years after GKR, respectively, and the
tumor size remained unchanged. In our experience with infe-
rior doses, we can obtain long-term control of the treated target
with no complications.
Management of Recurrent Tumor after Surgery
Continuous follow-up after radical removal is justified by
potential recurrence. The actual recurrence rate is difficult to
assess, even in the era of modern neuroimaging. This rate
may range between 6 and 39% of all cases (12, 16, 21, 24, 29).
The interval between radical removal and recurrence is usu-
ally short, with only one case reported at 4 years in the liter-
ature (29). Herman et al. (16) observed that no recurrence was
recorded after 24 months. In the 72 patients managed by
Lloyd et al. (21), 46.5% of recurrences presented within 1 year
and some within 6 months. In our own series, we clearly dis-
tinguished the situation of regrowth after subtotal resection
from true recurrences after GTR, which may explain why our
findings of later recurrence (2 and 4 yr after surgery) do not
match previous reports. Several studies have shown that the
incidence of recurrence correlates to the degree of cranial base
extension and the quality of initial resection (16, 17, 21).
Special interest has been brought to the involvement of the
basisphenoid. Howard et al. (17) compared the recurrence
rate in two consecutively operated cohorts of 20 patients each.
In the most recent group, additional surgical exploration of
the basisphenoid was conducted with exposition of the ante-
rior part of the pterygoid canal and drilling of the basisphe-
noid. In this latter group, no recurrence was observed,
VOLUME 61 | NUMBER 4 | OCTOBER 2007 | 775
ROCHE ET AL.
whereas it reached a 40% rate in the former group. In another
study, Lloyd et al. (21) indicated a 93% rate of recurrence in
the group of patients displaying invasion at the base of the
pterygoid process compared with only 7% in the group in
whom the sphenoid was not invaded. In our policy, we sys-
tematically treated the four cases of recurrence using a new
surgical procedure with long-term cure.
Tumor Classifications and Current Management
Several classifications of JNA extension (3, 4, 13, 27) have
been proposed, all of which were described before the advent of
modern imaging and were not based on the regular pattern of
tumor spread. Moreover, some degrees of overlapping between
the different stages were observed. In the Fisch grading sys-
tem, Stage IV tumors are defined by an intracranial intradural
extension. Cavernous sinus, pituitary fossa, or optic chiasm
involvement is included in the “IV b” tumors because we con-
sider that cavernous sinus is not intradural and that this fre-
quent involvement cannot be compared with the rare optic
pathway involvement.
For each individual case, we suggest describing the exact
pattern of the tumor extensions and then adopting a tailored
surgery following simple rules. Focal JNAs may be approached
using noninvasive maxillofacial routes. JNAs that display a
discrete involvement to the cranial base (pterygopalatine fossa,
pterygoid plate, inferior compartment of the orbit) may be still
operated on by regular maxillofacial approaches. Tumors
spreading extensively to the cranial base or reaching the
intracranial extracavernous regions are good candidates for
neurosurgical cranial base techniques or for combined
approaches. Large JNAs that display intracavernous involve-
ment require extracavernous surgery and adjunctive radiation
technique for the intracavernous remnant tumor.
CONCLUSIONS
Invasive JNAs remain challenging because of the difficulties
eradicating the multiple extensions of the tumors to the cranial
base and intracranial structures. A thorough preoperative radi-
ological workup of these extensions is required to plan an indi-
vidual tailored treatment. The satisfactory clinical results
obtained from the present series strongly support the concept
of a multimodality management when preoperative emboliza-
tion is required and collaborative maxillofacial and neurosurgi-
cal management offer the best chances for optimal resection. In
the case of remnant or recurrent tumors, a second surgery is
considered as the best option if the tumor can be reasonably
extirpated. Involvement of the cavernous sinus does not
require aggressive surgery and can be advantageously treated
with a radiosurgical procedure.
REFERENCES
1. Andrews JC, Fisch U, Valavanis A, Aeppli U, Makek M: The surgical manage-
ment of extensive nasopharyngeal angiofibromas with the infratemporal
fossa approach. Laryngoscope 99:429–437, 1989.
776 | VOLUME 61 | NUMBER 4 | OCTOBER 2007
2. Briant TD, Fitzpatrick PJ, Book H: The radiological treatment of juvenile
nasopharyngeal angiofibromas. Ann Otol Rhinol Laryngol 79:1108–1113,
1970.
3. Chandler JR, Goulding R, Moskowitz L, Quencer RM: Nasopharyngeal
angiofibroma: Staging and management. Ann Otol Rhinol Laryngol
93:322–329, 1984.
4. Chaveau C: History of the diseases from the pharynx [in French]. Paris, Ballière,
1906, p 562.
5. Close LG, Schaefer SD, Mickey BE, Manning SC: Surgical management of
nasopharyngeal angiofibroma involving the cavernous sinus. Arch Otolaryn-
gol Head Neck Surg 115:1091–1095, 1989.
6. Cummings BJ, Blend R, Keane T, Fitzpatrick P, Beale F, Clark R, Garrett P,
Harwood A, Payne D, Rider W: Primary radiation therapy for juvenile
nasopharyngeal angiofibroma. Laryngoscope 94:1599–1605, 1984.
7. Dare AO, Gibbons KJ, Proulx GM, Fenstermaker RA: Resection followed by
radiosurgery for advanced juvenile nasopharyngeal angiofibroma: Report of
two cases. Neurosurgery, 52:1207–1211, 2003.
8. Di Rienzo A, Ricci A, Scogna A, Zotta D, Stati G, Galzio R, Tschabitscher M:
The open-mouth fronto-orbitotemporozygomatic approach for extensive
benign tumors with coexisting splanchnocranial and neurocranial involve-
ment. Neurosurgery 54:1170–1180, 2004.
9. Dohar JE, Duvall AJ 3rd: Spontaneous regression of juvenile nasopharyngeal
angiofibroma. Ann Otol Rhinol Laryngol 101:469–471, 1992.
10. Economou TS, Abemayor E, Ward PH: Juvenile nasopharyngeal angiofi-
broma: An update of the UCLA experience, 1960–1985. Laryngoscope
98:170–175, 1988.
11. English GM, Hemenway WG, Cundy RL: Surgical treatment of invasive
angiofibroma. Arch Otolaryngol 96:312–318, 1972.
12. Fagan JJ, Snyderman CH, Carrau RL, Janecka IP: Nasopharyngeal angiofibro-
mas: Selecting a surgical approach. Head Neck 19:391–399, 1997.
13. Fisch U: The infratemporal fossa approach for nasopharyngeal tumors.
Laryngoscope 93:36–44, 1983.
14. Fisch U, Mattox D: Microsurgery of the Skull Base. New York, Thieme Medical
Publishers, 1988, p 350.
15. Hakuba A, Liu S, Nishimura S: The orbitozygomatic infratemporal approach:
A new surgical approach. Surg Neurol 26:271–276, 1986.
16. Herman P, Lot G, Chapot R, Salvan D, Huy PT: Long-term follow-up of juve-
nile nasopharyngeal angiofibromas: Analysis of recurrences. Laryngoscope
109:140–147, 1999.
17. Howard DJ, Lloyd G, Lund V: Recurrence and its avoidance in juvenile
angiofibroma. Laryngoscope 111:1509–1511, 2001.
18. Jafek BW, Krekorian EA, Kirsch WM, Wood RP: Juvenile nasopharyngeal
angiofibroma: Management of intracranial extension. Head Neck Surg
2:119–128, 1979.
19. Janecka IP: Classification of facial translocation approach to the skull base.
Otolaryngol Head Neck Surg 112:579–585, 1995.
20. Jones GC, DeSanto LW, Bremer JW, Neel HB 3rd: Juvenile angiofibromas.
Behavior and treatment of extensive and residual tumors. Arch Otolaryngol
Head Neck Surg 112:1191–1193, 1986.
21. Lloyd G, Howard D, Phelps P, Cheesman A: Juvenile angiofibroma: The les-
sons of 20 years of modern imaging. J Laryngol Otol 113:127–134, 1999.
22. McCombe A, Lund VJ, Howard DJ: Recurrence in juvenile angiofibroma.
Rhinology 28:97–102, 1990.
23. Pryor SG, Moore EJ, Kasperbauer JL: Endoscopic versus traditional approaches
for excision of juvenile nasopharyngeal angiofibroma. Laryngoscope
115:1201–1207, 2005.
24. Radkowski D, McGill T, Healy GB, Ohlms L, Jones DT: Angiofibroma.
Changes in staging and treatment. Arch Otolaryngol Head Neck Surg
122:122–129, 1996.
25. Reddy KA, Mendenhall WM, Amdur RJ, Stringer SP, Cassisi NJ: Long-term
results of radiation therapy for juvenile nasopharyngeal angiofibroma. Am J
Otolaryngol 22:172–175, 2001.
26. Sekhar LN, Schramm VL Jr, Jones NF: Subtemporal-preauricular infratempo-
ral fossa approach to large lateral and posterior cranial base neoplasms.
J Neurosurg 67:488–499, 1987.
27. Sessions RB, Bryan RN, Naclerio RM, Alford BR: Radiographic staging of
juvenile angiofibroma. Head Neck Surg 3:279–283, 1981.
www.neurosurgery-online.com

28. Standefer J, Holt GR, Brown WE Jr, Gates GA: Combined intracranial and
extracranial excision of nasopharyngeal angiofibroma. Laryngoscope
93:772–779, 1983.
29. Zhang M, Garvis W, Linder T, Fisch U: Update on the infratemporal fossa
approaches to nasopharyngeal angiofibroma. Laryngoscope 108:1717–1723,
1998.
COMMENTS
I n our experience with these large extensive lesions, we too have
found that routine preoperative embolization helps with resection
(1, 5). However, our approach is slightly different from that of the
authors. We prefer anterior craniofacial resection via a modified trans-
basal approach as a craniotomy-only type of craniofacial resection.
During surgery, if a radical resection is considered infeasible, a sub-
labial transfacial approach, consisting mostly of a medial maxillectomy
or a LeFort I osteotomy, is added (2, 4). We rarely use lateral cranial
base approaches for these extensive tumors. We fully agree with the
authors and do not resect tumoral components that invade the cav-
ernous sinus. They are either radiated postoperatively or later if the
tumor progresses.
Using our approach, we have not encountered any significant cos-
metic deformities because facial incisions are avoided (2). Facial numb-
ness is also avoided during a lateral approach. Furthermore, anosmia
is effectively avoided by the addition of a cribriform plate osteotomy.
Because these patients' tumors are benign, they are perfect candidates
for an olfactory-preserving procedure (3). Using anterior craniofacial
resection for these extensive tumors, our rate of recurrence or incom-
plete resection is no higher than commonly reported in contemporary
series.
We believe that a craniofacial approach is key. As the authors point
out, a significant number of these tumors receive their blood supply
directly from the cavernous portion of the internal carotid artery. These
feeders are best managed via an anterior transbasal procedure rather
than via a transfacial approach. Potentially with the latter, a small
amount of residual tumor could be left.
Iman Feiz-Erfan
Robert F. Spetzler
Phoenix, Arizona
1. Deshmukh VR, Fiorella DJ, McDougall CG, Spetzler RF, Albuquerque FC:
Preoperative embolization of central nervous system tumors. Neurosurg Clin
N Am 16:411–432, 2005.
2. Feiz-Erfan I, Han PP, Spetzler RF, Porter RW, Klopfenstein JD, Ferreira MA,
Beals SP, Joganic EF: Exposure of midline cranial base without a facial incision
through a combined craniofacial-transfacial procedure. Neurosurgery 56
[Suppl 1]:28–35, 2005.
3. Feiz-Erfan I, Han PP, Spetzler RF, Horn EM, Klopfenstein JD, Kim LJ, Porter
RW, Beals SP, Lettieri SC, Joganic EF: Preserving olfactory function in anterior
craniofacial surgery through cribriform plate osteotomy applied in selected
patients. Neurosurgery 57 [Suppl 1]:86–93, 2005.
4. Feiz-Erfan I, Han PP, Spetzler RF, Horn EM, Klopfenstein JD, Porter RW,
Ferreira MA, Beals SP, Lettieri SC, Joganic EF: The radical transbasal approach
for resection of anterior and midline skull base lesions. J Neurosurg
103:485–490, 2005.
5. Kim LJ, Albuquerque FC, Aziz-Sultan A, Spetzler RF, McDougall CG: Low
morbidity associated with use of n-butyl cyanoacrylate liquid adhesive for pre-
operative transarterial embolization of central nervous system tumors.
Neurosurgery 59:98–104, 2006.
NEUROSURGERY
INVASIVE JUVENILE NASOPHARYNGEAL ANGIOFIBROMAS
I n this very elegant series, the authors have described the multimodal-
ity treatment of extensive juvenile nasopharyngeal angiofibromas
(JNAs). I agree with the authors that extensive intracranial extensions
can be removed by an infratemporal approach, without the manipula-
tion of the temporomandibular joint. Preoperative embolization is very
important in this group of patients because it can convert a very vas-
cular tumor to one that is not very vascular. In addition to embolization
through the external carotid artery branches, we have sometimes used
direct puncture of the tumor and embolization through this route.
When the cavernous sinus is involved by this tumor, this piece is very
easily removed from the neurovascular structures. I would, therefore,
encourage the authors to pursue the remnants inside the cavernous
sinus more aggressively. Cosmetic considerations are very important
for these patients because many of them are in the developmental stage
when these tumors are removed. Some of the patients may require sec-
ondary reconstruction. Extensive tumor devascularization before sur-
gery may allow a less invasive approach and a better cosmetic outcome
in the long term. The authors are commended for bringing this exten-
sive and carefully studied series to the neurosurgical literature.
Laligam N. Sekhar
Seattle, Washington
T his report by Roche et al. highlights the issues surrounding the
management of extensive JNAs. These tumors typically occur in
adolescent males and usually present with nasal obstruction, epistaxis,
and Eustachian tube dysfunction. Delay in presentation is common
given the tendency to associate the indolent symptoms of JNA with
common sinonasal inflammatory disease. Thus, many tumors are
advanced at diagnosis, with intracranial and infratemporal fossa exten-
sion being common. Endovascular embolization of the tumor typically
precedes the judicial application and often a combination of varied
approaches to the sinonasal cavities, pterygopalatine and pterygomax-
illary fissures, infratemporal fossa, and cranial base, including the cav-
ernous sinus and parasellar region. This allows for the safe and com-
plete removal of the vast majority of extensive tumors such as those
described by the authors. Combining anterior approaches with the
epidural temporopolar or modified subtemporal-infratemporal
approaches, as described by the authors, is optimal and can access
common sites of residual tumor. There is absolutely no need to induce
a conductive hearing loss by using any of the postauricular approaches
that have been described for the resection of these tumors. I completely
agree with the authors that patients with unexpected residual or recur-
rence that is deemed resectable should be offered further surgery to
achieve complete tumor removal for these histologically benign
tumors. In my opinion, the uniform application of postoperative irra-
diation in the setting of “unresectable” residual tumor in this group of
young individuals is unwarranted given the uncertain biological nature
of the tumor. Many residual tumors remain indolent and some, in fact,
have been shown to spontaneously regress. The use of radiosurgery,
fractionated stereotactic radiotherapy, or intensity-modulated radia-
tion therapy should be reserved for instances of proven tumor growth
given our ability to carefully monitor these individuals with magnetic
resonance imaging scans, the uncertainty of tumor biology, and the
young age of these patients. Both hormonal and chemotherapeutic
treatments have been reported for JNAs, but their success remains
anecdotal.
Franco DeMonte
Houston, Texas
VOLUME 61 | NUMBER 4 | OCTOBER 2007 | 777