- Email: [email protected]
Juvenile nasopharyngeal angiofibroma: Evaluation and surgical management of advanced disease
Otolaryngology–Head and Neck Surgery (2008) 138, 581-586
ORIGINAL RESEARCH—SKULL BASE SURGERY
Juvenile nasopharyngeal angiofibroma: Evaluation and surgical management of advanced disease Giovanni Danesi, MD, Davide T. Panciera, MD, Richard J. Harvey, MBBS, FRACS, and Cristina Agostinis, MD, Bergamo, Italy OBJECTIVE: Evaluate preoperative imaging in predicting operative stage. Describe the outcomes in surgically treated juvenile nasopharyngeal angiofibroma (JNA) with the influence of middle cranial fossa, carotid, or dural involvement on recurrence. STUDY DESIGN: Retrospective cohort of surgically treated patients with JNA. SUBJECTS AND METHODS: Eighty-five patients from a regional Italian referral center were assessed for recurrence, radiologic, and operative staging. High risk areas involved were recorded at surgery. RESULTS: Recurrence for advanced disease (IIIb⫹) was 18.2% (6 of 33) and 15.3% (13 of 85) overall. Preoperative staging poorly correlated with operative stage (P ⫽ 0.15). No single high risk area was predictive for recurrence, but the absence of any risk factor was associated with a favorable outcome (P ⬍ 0.01). CONCLUSION: Dural involvement by tumor is rare and imaging may overstage disease. Anterior access, endoscopic or open, is sufficient to address intracranial involvement. When an open approach is used, a midface degloving technique affords excellent exposure even for advanced disease. Lateral approaches with their associated morbidity can be reserved for selected recurrent disease. © 2008 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved.
aforementioned routes was considered inoperable. Chemotherapy and radiotherapy were considered the only viable treatment strategies.3 Complications from radiation therapy for intracranial JNA are known to induce sarcomas, temporal lobe necrosis, cranial nerve palsies, hypopituitarism, and growth arrest. These potential adverse effects of irradiation cannot be ignored in a group of young patients such as those with JNA.3 In many centers, chemotherapy has been largely abandoned due to associated side effects and is rarely used.3 Surgical excision is consider by many to be the primary treatment modality4 and is generally preferred to radiotherapy as initial treatment for JNA with intracranial extension. The development of preoperative tumor embolization and surgical refinement in anterolateral skull base surgery underlies this approach.5 Preoperative evaluation of intracranial tumor involvement may significantly influence a surgical approach and its associated morbidity. This article discusses the ability to accurately predict the operative stage from imaging in advanced cases and the outcomes of extensive JNA compared with lower staged disease. The influence of middle cranial fossa (MCF), carotid, or true dural involvement on recurrence is also discussed.
A
METHODS
lthough histologically benign, juvenile nasopharyngeal angiofibroma (JNA) can cause significant morbidity and occasional mortality through aggressive submucosal spread to adjacent structures.1 Intracranial extension of JNA may occur with tumor spreading through well described routes to the skull base.2 This spread is dominated by an extradural pattern.1,2 Many routes to the cranium are available from the tumor’s point of origin, near the sphenopalatine foramen. Tumor extension can occur to the infratemporal fossa through the pterygomaxillary space and onto the middle cranial fossa via the superior or inferior orbital fissures. Spread may also occur into the anterior cranial fossa. Superior extension directly via the roof of the sphenoid sinus results in tumor medial to the internal carotid artery and lateral to the pituitary gland. Direct superior extension through the cribriform plate is uncommon.2 Until recently, intracranial extension of JNA, by some of the
Between 1981 and 2004, 85 patients with JNAs were operated on in the Department of Otolaryngology–Head and Neck Surgery at a regional referral center in Italy. All patients were males and age ranged from 11 to 33 years. The most common symptoms of presentation were nasal obstruction, severe epistaxis, and facial swelling. The duration of symptoms until diagnosis ranged from 1 to 14 months. Demographic data are presented in Table 1. The study was approved by the ethics commission of Ospedali Riuniti di Bergamo.
Assessment: Presurgery and Postsurgery All patients were studied with computed tomography (CT) ⫾ magnetic resonance imaging (MRI) ⫾ carotid angiogra-
Received June 13, 2007; revised October 5, 2007; accepted January 23, 2008.
0194-5998/$34.00 © 2008 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2008.01.011
582
Otolaryngology–Head and Neck Surgery, Vol 138, No 5, May 2008
Table 1 Demographic data for the intracranial and total group Intracranial extension (Fisch IIIB⫹)
All JNA
Number 33 85 Age (years mean ⫾ 16.4 ⫾ 4.0 17.1 ⫾ 4.2 SD) Range (years) 11-24 11-33 Gender (male) 100% 100% Follow- up (months 51.4 ⫾ 26.1 54.9 ⫾ 25.6 mean ⫾ SD) Recurrence 18.2% (6 of 33) 15.3% (13 of 85)
phy, and the tumors were staged according to the Andrew’s modification of the classification of Fisch6 (Table 2). CT and MRI were performed in all cases where intracranial extension was suspected (stage IIIb⫹). A table of cases by CT/MRI stage is presented in Table 3. During surgery, risk factors were recorded as true dural involvement, tumor adjacent to MCF foramen or dura, next to the ICA, and dissection from the cavernous sinus. A single tumor could be assigned to multiple risk groups. All IIIb⫹ cases were studied with MRI ⫾ carotid angiography between 6 and 12 months postoperatively. Interval CT or MRI was then used for further follow-up every 6 months for 2 years and then yearly.
Preoperative Management Preoperative angiography was performed on all IIIb⫹ lesions. Vascular supply in all tumors included the distal branches of internal maxillary artery. Internal carotid artery (ICA) vascular contribution was shown in some tumors with intracranial extension. Routine embolization was only performed on all stage IIIb⫹ before surgery, In 2 cases of significant ICA involvement, direct intratumoral embolization with temporary balloon occlusion of the ICA was undertaken.
Surgical Management Primary surgery was performed via an anterior approach in all cases. A lateral rhinotomy with malar-cheek-maxillary
flap was performed in 2 cases and the remaining cases with a nasomaxillary pedicled flap. All cases of midface degloving included temporary removal of the front face of the maxilla for operative access. Surgical approach by stage is presented in Table 3. Extension of disease was noted during the operation in all cases. Dural involvement was defined as true transdural extension of tumor where no surgical plane could be found. In these cases, the dura was resected with tumor and reconstruction undertaken. Surgical biopsy was performed and histology confirmed JNA on all 85 cases. We acknowledge that large tumors up to IIIa disease (and potentially larger) can be successfully managed with an endoscopic technique.7 An endoscopic approach for lower stage disease was not undertaken at our institution because of the lack of endoscopic skull base experience.
Statistical Analysis Statistical analysis was performed using the Statistical Package for the Social Sciences software (SPSS; Version 15.0; SPSS Inc., Chicago, IL). Bivariate correlations were calculated with Spearman correlation coefficents for ordinal values. Chi squared and Fisher exact statistics were used for recurrence data and risk groups.
RESULTS Previous experience had highlighted that MRI tended to “overstage” the advanced tumors and poorly define dural involvement.8 Frequently, a surgical plane could be found adjacent to the dura and ICA. A scatter plot of preoperative stage, as determined by CT and MRI is compared with the operative staging of the tumor in Figure 1. Overall preoperative stage to operative stage correlated well with a Spearman’s coefficient of 0.895 (SE 0.25) (P ⬍ 0.001). This corresponded to Kendalls tau-b 0.82 (P ⬍ 0.01) when treated as ordinal data. However, for the IIIb⫹ group, there was a general overstaging and poor association (Table 4). The overall recurrence rate was 15.3% (13 of 85). We had residual tumor in 6 (18.2%) of 34 patients with JNA with intracranial extension; this was not statistically different to the extracranial group recurrence rate of 13.5% (7 of 52) (P ⫽ 0.55). Revision surgery was performed for per-
Table 2 Andrew’s (modified Fisch) classification Type I II IIIa IIIb IVa IVb
Tumor extent Limited to the nasopharynx and nasal cavity, bone destruction negligible, or limited to the sphenopalatine foramen Invades the pterygopalatine fossa or the maxillary, ethmoid or sphenoid sinus with bone destruction Invades the infratemporal fossa or the orbital region without intracranial involvement Invades the infratemporal fossa or orbit with intracranial extradural (parasellar) involvement Intracranial intradural tumor without infiltration of the cavernous sinus, pituitary fossa, or optic chiasm Intracranial intradural tumor with infiltration of the cavernous sinus, pituitary fossa, or optic chiasm
Danesi et al
Juvenile nasopharyngeal angiofibroma Evaluation and . . .
Table 3 Operative approach by stage CT/MRI based Lateral Midface stage rhinotomy degloving I II IIIa IIIb Iva Ivb
5 19 7 7 3 0
4 8 9 17 5 1
Total number of cases 9 27 16 24 8 1
sistent disease in all of the 13 cases. Total gross surgical resection was attempted in all but one case during the first operation. For this patient, the intracranial part of the tumor penetrated the dura of middle cranial fossa. The transdural part of the tumor was later removed with a lateral skull base approach (subtemporal) by the same surgeon as a planned procedure. Residual tumor was identified on angiography in 4 patients with intracranial extension; in 1 case, the residual tumor was identified lateral to the cavernous sinus and for removal a subtemporal approach was performed. One patient presented with residual tumor in the middle cranial fossa floor (extradural), and a subtemporal approach was performed. Midface degloving was performed for tumor found in pterygomaxillary fossa and infratemporal fossa and also for residual tumor found in the nasopharynx. In another case, focal residual tumor in pterygomaxillary fossa was removed with an endoscopic approach. The large sinonasal cavity created by previous surgery afforded the opportunity
583
Table 4 Preoperative vs operative staging: the disparity of accurate staging with advanced disease Staging by Number CT/MRI of cases I - IIIa IIIb⫹
52 33
Operative correlation (Spearman’s coefficent) P value 0.78 0.26
⬍0.01* 0.15
to endoscopically remove focal residual tumor. Overall, 6 lateral rhinotomy cases were revised by the same approach and 1 by the midface. Conversion to a midface degloving approach, lateral skullbase resection, and endoscopic removal was required for 2 patients in each group. In the recurrent group, only 1 had alar rim necrosis and 2 had asymptomatic ICA thrombosis discovered during surgical removal of recurrence. The identification of a surgical plane between dura and tumor was a consisting finding previously reported.8 An analysis of recurrence by risk factor demonstrated that no single anatomic area predicted recurrence. As expected, absence of any risk factor was a negative predictor for recurrence. The association of recurrence to risk factor is presented in Table 5. For 2 cases of JNA with vascular supply from the ICA, intratumor embolization with glue and temporary ballooning occlusion of the ICA was used. Anecdotally, a reduction of intraoperative bleeding was not seen in these cases. Subsequently, preoperative embolization is no longer used in this group due to poor efficacy and the potential neurologic risk associated with the intervention. Intraoperative bleeding did not prevent safe and radical removal of advanced tumor with an anterior approach.
DISCUSSION
Figure 1 Scatter plot of stage, as determined by preoperative imaging against pathologic findings at operation.
Surgical excision of JNA is considered the primary treatment modality, whether intracranial or extracranial.1 The majority of tumors may be successfully removed with an endoscopic approach.9 The open anterior approach may be chosen for those with intracranial extension. When intracranial extension occurs, the most common sites of intracranial extension of JNA include the middle cranial fossa, the pituitary fossa/parasellar, and the anterior cranial fossa.10 True invasion of the cavernous sinus is unusual.10 Preoperative radiologic evaluation for intracranial extension of the tumor is important in planning the surgical approach,11 in particular, for identifying structures such as the ICA, cavernous sinus, and dural involvement.11 Our basic assumption is that JNA is a benign tumor that, even if intracranial extension is present, rarely penetrates the dura.10 Ideally, preoperative radiology should demonstrate a plane between tumor and neurovascular structures.10 In our experience,
584
Otolaryngology–Head and Neck Surgery, Vol 138, No 5, May 2008
Table 5 Correlation of risk factor for recurrence High risk area involved at operation Dural involvement Cavernous sinus MCF Carotid No risk group (negative association)
Recurrence (total cases with risk) 0 5 6 1 5
(n (n (n (n (n
⫽ ⫽ ⫽ ⫽ ⫽
5) 18) 34) 4) 45)
Chi-squared (P value) 0.96 2.75 0.24 0.31 13.64
(0.33) (0.10) (0.62) (0.58) (0.001)
Fisher’s exact (P ⬍ 0.05 for significance) NS 0.14 0.76 0.49 0.001
combination CT and MRI has not been always able to achieve this. The ability to distinguish if a lesion is truly intradural preoperatively can be difficult. Absence of a cleavage plane between the tumor and dura or encasement (partial or total) of the ICA is a potential sign of a transdural lesion. MRI criteria to identify dural penetration are lacking in JNA. The transdural radiologic criteria for extracranial pathoses (ie, cordoma, paraganglioma, and JNA) are difficult to define. The presence of cerebral parenchymal vessels with collateral branches that feed the tumor may be a sign of dural penetration. In our series, no preoperative carotid angiography demonstrated a JNA was fed by parenchymal branches. CT and MRI may be more useful in identifying the interface between tumor and vascular structures. However, even with partial or total encasement of the ICA on preoperative MRI, a dural plane was usually found at operation. MRI is not always able to distinguish intracranial JNA as extra- or intradural.11 This often misguides preoperative surgical planning in these cases. The size and the route of intracranial invasion may influence the surgical choice.1,12 A tumor that extends intracranially and remains lateral to the cavernous sinus is accessible through either a lateral skull base or an anterior approach. When a tumor extends intracranially and remains medial or inferomedial to the cavernous sinus, it is more readily removed through the anterior approach.1 In our
center, all tumors with intracranial extension (33 of 85) were primarily removed through an anterior approach. Midface degloving is preferred for the wide exposure and avoidance of facial scarring. With this approach, it is possible to gain wide access to both nasal cavities, maxillary, ethmoid and sphenoid sinuses, pterygopalatine fossae, nasopharynx, and the infratemporal fossa.13 The endoscopic approach may be applicable to many of our cases, but an open approach was chosen due to lack of endoscopic skull base experience. The combined lateral and anterior approach may represent an alternative to a transfacial approach when the tumor extends intracranially and laterally.1 The lateral approach allows a wide access to the tumor with exposure of the middle cranial fossa floor, the carotid canal, and the lateral aspect of the cavernous sinus. Unfortunately, this approach is responsible for unilateral hearing loss, facial anaesthesia (secondary to V2 and V3 resection) and temporal depression due to temporalis muscle rotation.1 We prefer the anterior approach as an ideal first surgical option even when tumor extends intracranially and rests lateral to the cavernous sinus. In the literature, there are very few reports of JNA that extend both intracranial and intradural. The lateral approach should be limited, in our opinion, to selected intracranial recurrence. Recurrence or persistence is well reported in angiofibroma despite its benign histology and absence of multifo-
Figure 2 MRI scan of suspected intradural involvement preoperative but extradural at operation.
Figure 3 Suspected middle fossa and cavernous sinus involvement of JNA but the tumor remained extradural.
Danesi et al
Juvenile nasopharyngeal angiofibroma Evaluation and . . .
Figure 4 Carotid encasement and cavernous sinus involvement on MRI. Extradural surgical planes were available at surgery.
cal origin. Tumor recurrence is primarily seen with high stage tumors.1,2,14 The high rate of recurrence in tumor with intracranial extension may, in part, be due to a failure to recognize the clinical importance of skull base erosion.1,12 Recurrence rate for JNA varies up to 46% and often depends on surgical technique and tumor growth rate.15 More frequent recurrence correlates with extension to the sphenoid sinus, the base of pterygoid, and the clivus bone.15 For this reason, we now drill out what remains of pterygoid plates and the clivus bone (between the sphenoid sinus floor and midclivus). The vascularity of JNA is important in the planning of surgical resection.16,17 Blood supply is determined by the size and extension of the tumor. In the initial stages, there is a consistent vascular supply from the distal branches of external carotid artery. In advanced stage, a tumor may receive important vessels from the internal carotid artery, especially from collateral branches of the ophthalmic artery.18,19 Preoperative carotid angiography greatly assists in determining the vascular contributions to the tumor and to identify significant arteriovenous shunts.17-19 Even with high resolution angiography, only blood vessels with diameters exceeding 200 m can be visualized.17 Arteriovenous shunts of smaller calibre may not be identified with angiography and present a potential risk during embolization.17,18 JNA can have a high incidence of arteriovenous shunts and larger particles of PVA (’140 m in calibre) are recommend for preoperative embolization.5 Some authors argue for the need to perform direct intratumoral embolization in addition to preoperative embolization in cases of JNA with significant vascular contribution from the ICA.18 Temporary balloon occlusion of the ICA without neurologic complication was used in the preoperative management in 2 patients from our group. Little additional benefit was realized from direct intratumoral embolization with temporary balloon occlusion of ICA. Weighing in the potential complications and our brief experience, we do not believe this is necessary for the treatment of intracranial
585
JNA. A vascular supply from the ICA is not unusual for large tumors extending intracranially and does not limit complete and safe surgical excision.5,17,18 Comparable rates of complete removal of intracranial JNA to less aggressive disease were achieved. Radiologic follow-up is essential in the early identification of residual or recurrent disease.14,20 Postoperative MRI and contrast-enhanced CT have been recommended to reveal residual or recurrence tumor.21,22 Recurrence may also be demonstrated by carotid angiography or by subtraction MRI.11 In our institution, all patients have been studied with MRI ⫾ carotid angiography between 6 and 12 month postoperation. Subsequent follow-up is by CT or MRI yearly. Carotid angiography confirmed disease in 2 cases where MRI did not show recurrence. MRI, carotid angiography, and contrast-enhanced CT are all useful in the detection of residual or recurrent disease with little evidence to show the superiority of one modality.
CONCLUSION Preoperative imaging will often overstage advanced tumors. Advanced JNA can be successfully removed in 82% of patients during their first surgery despite intracranial extension. The majority of advanced JNA remain extradural, and a surgical plane can usually be found between important structures such as the ICA, cavernous sinus, and MCF. Where possible, an endoscopic approach maybe the first option for many patients. The endoscopic surgeon will also benefit from defined tumor-dural dissection planes. Anterior access, whether endoscopic or open, is often sufficient to address intracranial involvement. When an open anterior approach is used, the midface degloving technique affords excellent exposure even for advanced disease. Lateral approaches with their associated morbidity can be reserved for selected recurrent disease. Figures 2 to 4
AUTHOR INFORMATION From the Divisions of ORL and skull base surgery (Drs Danesi, Panciera, and Harvey) and Neuroradiology (Dr Agostinis), Ospedali Riuniti di Bergamo, Bergamo, Italy. Corresponding author: Richard Harvey, Flat 170, Green Place Flats, Dr Diogo de Faria, 1201, Vila Mariana, CEP 04037-004, Sao Paulo, SP, Brazil. E-mail address: [email protected].
AUTHOR CONTRIBUTIONS Giovanni Danesi, lead author, primary surgeon; Davide T. Panciera, surgical and clinical research participation; Richard J. Harvey, database, clinical research, statistics, manuscript preparation; Cristina Agostinis, radiologic expertise and interpretation.
586
Otolaryngology–Head and Neck Surgery, Vol 138, No 5, May 2008
FINANCIAL DISCLOSURE None.
REFERENCES 1. Bales C, Kotapka M, Loevner LA, et al. Craniofacial resection of advanced juvenile nasopharyngeal angiofibroma. Arch Otolaryngol Head Neck Surg 2002;128:1071–1078. 2. Radkowski D, McGill T, Healy GB, et al. Angiofibroma: changes in staging and treatment. Arch Otolaryngol Head Neck Surg 1996;122: 122–129. 3. Lee JT, Chen P, Safa A, et al. The role of radiation in the treatment of advanced juvenile angiofibroma. Laryngoscope 2002;112:1213–1220. 4. Marshall AH, Bradley PJ. Management dilemmas in the treatment and follow-up of advanced juvenile nasopharyngeal angiofibroma. ORL J Otorhinolaryngol Relat Spec 2006;68:273–278. 5. Schroth G, Haldemann AR, Mariani L, et al. Preoperative embolization of paragangliomas and angiofibromas: measurement of intratumoral arteriovenous shunts. Arch Otolaryngol Head Neck Surg 1996; 122:1320 –1325. 6. Andrews JC, Fisch U, Valavanis A, et al. The surgical management of extensive nasopharyngeal angiofibromas with the infratemporal fossa approach. Laryngoscope 1989;99:429 – 437. 7. Wormald PJ, Van Hasselt A. Endoscopic removal of juvenile angiofibromas. Otolaryngol Head Neck Surg 2003;129:684 – 691. 8. Danesi G, Panizza B, Mazzoni A, et al. Anterior approaches in juvenile nasopharyngeal angiofibromas with intracranial extension. Otolaryngol Head Neck Surg 2000;122:277–283. 9. Douglas R, Wormald P-J. Endoscopic surgery for juvenile nasopharyngeal angiofibroma: where are the limits? Curr 2006;14:1–5. 10. Close LG, Schaefer SD, Mickey BE, et al. Surgical management of nasopharyngeal angiofibroma involving the cavernous sinus. Arch Otolaryngol Head Neck Surg 1989;115:1091–1095.
11. Lloyd G, Howard D, Phelps P, et al. Juvenile angiofibroma: the lessons of 20 years of modern imaging. J Laryngol Otol 1999;113:127–134. 12. Donald PJ, Enepikedes D, Boggan J. Giant juvenile nasopharyngeal angiofibroma: management by skull-base surgery. Arch Otolaryngol Head Neck Surg 2004;130:882– 886. 13. Browne JD. The midfacial degloving procedure for nasal, sinus, and nasopharyngeal tumors. Otolaryngol Clin North Am 2001;34:1095– 1104. 14. Scholtz AW, Appenroth E, Kammen-Jolly K, et al. Juvenile nasopharyngeal angiofibroma: management and therapy. Laryngoscope 2001; 111:681– 687. 15. Fagan JJ, Snyderman CH, Carrau RL, et al. Nasopharyngeal angiofibromas: selecting a surgical approach. Head Neck 1997;19:391–399. 16. Laurent A, Wassef M, Chapot R, et al. Partition of calibrated tris-acryl gelatin microspheres in the arterial vasculature of embolized nasopharyngeal angiofibromas and paragangliomas. J Vasc Interv Radiol 2005; 16:507–513. 17. Petruson K, Rodriguez-Catarino M, Petruson B, et al. Juvenile nasopharyngeal angiofibroma: long-term results in preoperative embolized and non-embolized patients. Acta Otolaryngol (Stockh) 2002;122:96 – 100. 18. Lefkowitz M, Giannotta SL, Hieshima G, et al. Embolization of neurosurgical lesions involving the ophthalmic artery. Neurosurgery 1998;43:1298 –1303. 19. Casasco A, Houdart E, Biondi A, et al. Major complications of percutaneous embolization of skull-base tumors. AJNR Am J Neuroradiol 1999;20:179 –181. 20. Mann WJ, Jecker P, Amedee RG. Juvenile angiofibromas: changing surgical concept over the last 20 years. Laryngoscope 2004;114:291–3. 21. Howard DJ, Lloyd G, Lund V. Recurrence and its avoidance in juvenile angiofibroma. Laryngoscope 2001;111:1509 –1511. 22. Kania RE, Sauvaget E, Guichard J-P, et al. Early postoperative CT scanning for juvenile nasopharyngeal angiofibroma: detection of residual disease. AJNR Am J Neuroradiol 2005;26:82– 88.
ORIGINAL RESEARCH—SKULL BASE SURGERY
Juvenile nasopharyngeal angiofibroma: Evaluation and surgical management of advanced disease Giovanni Danesi, MD, Davide T. Panciera, MD, Richard J. Harvey, MBBS, FRACS, and Cristina Agostinis, MD, Bergamo, Italy OBJECTIVE: Evaluate preoperative imaging in predicting operative stage. Describe the outcomes in surgically treated juvenile nasopharyngeal angiofibroma (JNA) with the influence of middle cranial fossa, carotid, or dural involvement on recurrence. STUDY DESIGN: Retrospective cohort of surgically treated patients with JNA. SUBJECTS AND METHODS: Eighty-five patients from a regional Italian referral center were assessed for recurrence, radiologic, and operative staging. High risk areas involved were recorded at surgery. RESULTS: Recurrence for advanced disease (IIIb⫹) was 18.2% (6 of 33) and 15.3% (13 of 85) overall. Preoperative staging poorly correlated with operative stage (P ⫽ 0.15). No single high risk area was predictive for recurrence, but the absence of any risk factor was associated with a favorable outcome (P ⬍ 0.01). CONCLUSION: Dural involvement by tumor is rare and imaging may overstage disease. Anterior access, endoscopic or open, is sufficient to address intracranial involvement. When an open approach is used, a midface degloving technique affords excellent exposure even for advanced disease. Lateral approaches with their associated morbidity can be reserved for selected recurrent disease. © 2008 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved.
aforementioned routes was considered inoperable. Chemotherapy and radiotherapy were considered the only viable treatment strategies.3 Complications from radiation therapy for intracranial JNA are known to induce sarcomas, temporal lobe necrosis, cranial nerve palsies, hypopituitarism, and growth arrest. These potential adverse effects of irradiation cannot be ignored in a group of young patients such as those with JNA.3 In many centers, chemotherapy has been largely abandoned due to associated side effects and is rarely used.3 Surgical excision is consider by many to be the primary treatment modality4 and is generally preferred to radiotherapy as initial treatment for JNA with intracranial extension. The development of preoperative tumor embolization and surgical refinement in anterolateral skull base surgery underlies this approach.5 Preoperative evaluation of intracranial tumor involvement may significantly influence a surgical approach and its associated morbidity. This article discusses the ability to accurately predict the operative stage from imaging in advanced cases and the outcomes of extensive JNA compared with lower staged disease. The influence of middle cranial fossa (MCF), carotid, or true dural involvement on recurrence is also discussed.
A
METHODS
lthough histologically benign, juvenile nasopharyngeal angiofibroma (JNA) can cause significant morbidity and occasional mortality through aggressive submucosal spread to adjacent structures.1 Intracranial extension of JNA may occur with tumor spreading through well described routes to the skull base.2 This spread is dominated by an extradural pattern.1,2 Many routes to the cranium are available from the tumor’s point of origin, near the sphenopalatine foramen. Tumor extension can occur to the infratemporal fossa through the pterygomaxillary space and onto the middle cranial fossa via the superior or inferior orbital fissures. Spread may also occur into the anterior cranial fossa. Superior extension directly via the roof of the sphenoid sinus results in tumor medial to the internal carotid artery and lateral to the pituitary gland. Direct superior extension through the cribriform plate is uncommon.2 Until recently, intracranial extension of JNA, by some of the
Between 1981 and 2004, 85 patients with JNAs were operated on in the Department of Otolaryngology–Head and Neck Surgery at a regional referral center in Italy. All patients were males and age ranged from 11 to 33 years. The most common symptoms of presentation were nasal obstruction, severe epistaxis, and facial swelling. The duration of symptoms until diagnosis ranged from 1 to 14 months. Demographic data are presented in Table 1. The study was approved by the ethics commission of Ospedali Riuniti di Bergamo.
Assessment: Presurgery and Postsurgery All patients were studied with computed tomography (CT) ⫾ magnetic resonance imaging (MRI) ⫾ carotid angiogra-
Received June 13, 2007; revised October 5, 2007; accepted January 23, 2008.
0194-5998/$34.00 © 2008 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2008.01.011
582
Otolaryngology–Head and Neck Surgery, Vol 138, No 5, May 2008
Table 1 Demographic data for the intracranial and total group Intracranial extension (Fisch IIIB⫹)
All JNA
Number 33 85 Age (years mean ⫾ 16.4 ⫾ 4.0 17.1 ⫾ 4.2 SD) Range (years) 11-24 11-33 Gender (male) 100% 100% Follow- up (months 51.4 ⫾ 26.1 54.9 ⫾ 25.6 mean ⫾ SD) Recurrence 18.2% (6 of 33) 15.3% (13 of 85)
phy, and the tumors were staged according to the Andrew’s modification of the classification of Fisch6 (Table 2). CT and MRI were performed in all cases where intracranial extension was suspected (stage IIIb⫹). A table of cases by CT/MRI stage is presented in Table 3. During surgery, risk factors were recorded as true dural involvement, tumor adjacent to MCF foramen or dura, next to the ICA, and dissection from the cavernous sinus. A single tumor could be assigned to multiple risk groups. All IIIb⫹ cases were studied with MRI ⫾ carotid angiography between 6 and 12 months postoperatively. Interval CT or MRI was then used for further follow-up every 6 months for 2 years and then yearly.
Preoperative Management Preoperative angiography was performed on all IIIb⫹ lesions. Vascular supply in all tumors included the distal branches of internal maxillary artery. Internal carotid artery (ICA) vascular contribution was shown in some tumors with intracranial extension. Routine embolization was only performed on all stage IIIb⫹ before surgery, In 2 cases of significant ICA involvement, direct intratumoral embolization with temporary balloon occlusion of the ICA was undertaken.
Surgical Management Primary surgery was performed via an anterior approach in all cases. A lateral rhinotomy with malar-cheek-maxillary
flap was performed in 2 cases and the remaining cases with a nasomaxillary pedicled flap. All cases of midface degloving included temporary removal of the front face of the maxilla for operative access. Surgical approach by stage is presented in Table 3. Extension of disease was noted during the operation in all cases. Dural involvement was defined as true transdural extension of tumor where no surgical plane could be found. In these cases, the dura was resected with tumor and reconstruction undertaken. Surgical biopsy was performed and histology confirmed JNA on all 85 cases. We acknowledge that large tumors up to IIIa disease (and potentially larger) can be successfully managed with an endoscopic technique.7 An endoscopic approach for lower stage disease was not undertaken at our institution because of the lack of endoscopic skull base experience.
Statistical Analysis Statistical analysis was performed using the Statistical Package for the Social Sciences software (SPSS; Version 15.0; SPSS Inc., Chicago, IL). Bivariate correlations were calculated with Spearman correlation coefficents for ordinal values. Chi squared and Fisher exact statistics were used for recurrence data and risk groups.
RESULTS Previous experience had highlighted that MRI tended to “overstage” the advanced tumors and poorly define dural involvement.8 Frequently, a surgical plane could be found adjacent to the dura and ICA. A scatter plot of preoperative stage, as determined by CT and MRI is compared with the operative staging of the tumor in Figure 1. Overall preoperative stage to operative stage correlated well with a Spearman’s coefficient of 0.895 (SE 0.25) (P ⬍ 0.001). This corresponded to Kendalls tau-b 0.82 (P ⬍ 0.01) when treated as ordinal data. However, for the IIIb⫹ group, there was a general overstaging and poor association (Table 4). The overall recurrence rate was 15.3% (13 of 85). We had residual tumor in 6 (18.2%) of 34 patients with JNA with intracranial extension; this was not statistically different to the extracranial group recurrence rate of 13.5% (7 of 52) (P ⫽ 0.55). Revision surgery was performed for per-
Table 2 Andrew’s (modified Fisch) classification Type I II IIIa IIIb IVa IVb
Tumor extent Limited to the nasopharynx and nasal cavity, bone destruction negligible, or limited to the sphenopalatine foramen Invades the pterygopalatine fossa or the maxillary, ethmoid or sphenoid sinus with bone destruction Invades the infratemporal fossa or the orbital region without intracranial involvement Invades the infratemporal fossa or orbit with intracranial extradural (parasellar) involvement Intracranial intradural tumor without infiltration of the cavernous sinus, pituitary fossa, or optic chiasm Intracranial intradural tumor with infiltration of the cavernous sinus, pituitary fossa, or optic chiasm
Danesi et al
Juvenile nasopharyngeal angiofibroma Evaluation and . . .
Table 3 Operative approach by stage CT/MRI based Lateral Midface stage rhinotomy degloving I II IIIa IIIb Iva Ivb
5 19 7 7 3 0
4 8 9 17 5 1
Total number of cases 9 27 16 24 8 1
sistent disease in all of the 13 cases. Total gross surgical resection was attempted in all but one case during the first operation. For this patient, the intracranial part of the tumor penetrated the dura of middle cranial fossa. The transdural part of the tumor was later removed with a lateral skull base approach (subtemporal) by the same surgeon as a planned procedure. Residual tumor was identified on angiography in 4 patients with intracranial extension; in 1 case, the residual tumor was identified lateral to the cavernous sinus and for removal a subtemporal approach was performed. One patient presented with residual tumor in the middle cranial fossa floor (extradural), and a subtemporal approach was performed. Midface degloving was performed for tumor found in pterygomaxillary fossa and infratemporal fossa and also for residual tumor found in the nasopharynx. In another case, focal residual tumor in pterygomaxillary fossa was removed with an endoscopic approach. The large sinonasal cavity created by previous surgery afforded the opportunity
583
Table 4 Preoperative vs operative staging: the disparity of accurate staging with advanced disease Staging by Number CT/MRI of cases I - IIIa IIIb⫹
52 33
Operative correlation (Spearman’s coefficent) P value 0.78 0.26
⬍0.01* 0.15
to endoscopically remove focal residual tumor. Overall, 6 lateral rhinotomy cases were revised by the same approach and 1 by the midface. Conversion to a midface degloving approach, lateral skullbase resection, and endoscopic removal was required for 2 patients in each group. In the recurrent group, only 1 had alar rim necrosis and 2 had asymptomatic ICA thrombosis discovered during surgical removal of recurrence. The identification of a surgical plane between dura and tumor was a consisting finding previously reported.8 An analysis of recurrence by risk factor demonstrated that no single anatomic area predicted recurrence. As expected, absence of any risk factor was a negative predictor for recurrence. The association of recurrence to risk factor is presented in Table 5. For 2 cases of JNA with vascular supply from the ICA, intratumor embolization with glue and temporary ballooning occlusion of the ICA was used. Anecdotally, a reduction of intraoperative bleeding was not seen in these cases. Subsequently, preoperative embolization is no longer used in this group due to poor efficacy and the potential neurologic risk associated with the intervention. Intraoperative bleeding did not prevent safe and radical removal of advanced tumor with an anterior approach.
DISCUSSION
Figure 1 Scatter plot of stage, as determined by preoperative imaging against pathologic findings at operation.
Surgical excision of JNA is considered the primary treatment modality, whether intracranial or extracranial.1 The majority of tumors may be successfully removed with an endoscopic approach.9 The open anterior approach may be chosen for those with intracranial extension. When intracranial extension occurs, the most common sites of intracranial extension of JNA include the middle cranial fossa, the pituitary fossa/parasellar, and the anterior cranial fossa.10 True invasion of the cavernous sinus is unusual.10 Preoperative radiologic evaluation for intracranial extension of the tumor is important in planning the surgical approach,11 in particular, for identifying structures such as the ICA, cavernous sinus, and dural involvement.11 Our basic assumption is that JNA is a benign tumor that, even if intracranial extension is present, rarely penetrates the dura.10 Ideally, preoperative radiology should demonstrate a plane between tumor and neurovascular structures.10 In our experience,
584
Otolaryngology–Head and Neck Surgery, Vol 138, No 5, May 2008
Table 5 Correlation of risk factor for recurrence High risk area involved at operation Dural involvement Cavernous sinus MCF Carotid No risk group (negative association)
Recurrence (total cases with risk) 0 5 6 1 5
(n (n (n (n (n
⫽ ⫽ ⫽ ⫽ ⫽
5) 18) 34) 4) 45)
Chi-squared (P value) 0.96 2.75 0.24 0.31 13.64
(0.33) (0.10) (0.62) (0.58) (0.001)
Fisher’s exact (P ⬍ 0.05 for significance) NS 0.14 0.76 0.49 0.001
combination CT and MRI has not been always able to achieve this. The ability to distinguish if a lesion is truly intradural preoperatively can be difficult. Absence of a cleavage plane between the tumor and dura or encasement (partial or total) of the ICA is a potential sign of a transdural lesion. MRI criteria to identify dural penetration are lacking in JNA. The transdural radiologic criteria for extracranial pathoses (ie, cordoma, paraganglioma, and JNA) are difficult to define. The presence of cerebral parenchymal vessels with collateral branches that feed the tumor may be a sign of dural penetration. In our series, no preoperative carotid angiography demonstrated a JNA was fed by parenchymal branches. CT and MRI may be more useful in identifying the interface between tumor and vascular structures. However, even with partial or total encasement of the ICA on preoperative MRI, a dural plane was usually found at operation. MRI is not always able to distinguish intracranial JNA as extra- or intradural.11 This often misguides preoperative surgical planning in these cases. The size and the route of intracranial invasion may influence the surgical choice.1,12 A tumor that extends intracranially and remains lateral to the cavernous sinus is accessible through either a lateral skull base or an anterior approach. When a tumor extends intracranially and remains medial or inferomedial to the cavernous sinus, it is more readily removed through the anterior approach.1 In our
center, all tumors with intracranial extension (33 of 85) were primarily removed through an anterior approach. Midface degloving is preferred for the wide exposure and avoidance of facial scarring. With this approach, it is possible to gain wide access to both nasal cavities, maxillary, ethmoid and sphenoid sinuses, pterygopalatine fossae, nasopharynx, and the infratemporal fossa.13 The endoscopic approach may be applicable to many of our cases, but an open approach was chosen due to lack of endoscopic skull base experience. The combined lateral and anterior approach may represent an alternative to a transfacial approach when the tumor extends intracranially and laterally.1 The lateral approach allows a wide access to the tumor with exposure of the middle cranial fossa floor, the carotid canal, and the lateral aspect of the cavernous sinus. Unfortunately, this approach is responsible for unilateral hearing loss, facial anaesthesia (secondary to V2 and V3 resection) and temporal depression due to temporalis muscle rotation.1 We prefer the anterior approach as an ideal first surgical option even when tumor extends intracranially and rests lateral to the cavernous sinus. In the literature, there are very few reports of JNA that extend both intracranial and intradural. The lateral approach should be limited, in our opinion, to selected intracranial recurrence. Recurrence or persistence is well reported in angiofibroma despite its benign histology and absence of multifo-
Figure 2 MRI scan of suspected intradural involvement preoperative but extradural at operation.
Figure 3 Suspected middle fossa and cavernous sinus involvement of JNA but the tumor remained extradural.
Danesi et al
Juvenile nasopharyngeal angiofibroma Evaluation and . . .
Figure 4 Carotid encasement and cavernous sinus involvement on MRI. Extradural surgical planes were available at surgery.
cal origin. Tumor recurrence is primarily seen with high stage tumors.1,2,14 The high rate of recurrence in tumor with intracranial extension may, in part, be due to a failure to recognize the clinical importance of skull base erosion.1,12 Recurrence rate for JNA varies up to 46% and often depends on surgical technique and tumor growth rate.15 More frequent recurrence correlates with extension to the sphenoid sinus, the base of pterygoid, and the clivus bone.15 For this reason, we now drill out what remains of pterygoid plates and the clivus bone (between the sphenoid sinus floor and midclivus). The vascularity of JNA is important in the planning of surgical resection.16,17 Blood supply is determined by the size and extension of the tumor. In the initial stages, there is a consistent vascular supply from the distal branches of external carotid artery. In advanced stage, a tumor may receive important vessels from the internal carotid artery, especially from collateral branches of the ophthalmic artery.18,19 Preoperative carotid angiography greatly assists in determining the vascular contributions to the tumor and to identify significant arteriovenous shunts.17-19 Even with high resolution angiography, only blood vessels with diameters exceeding 200 m can be visualized.17 Arteriovenous shunts of smaller calibre may not be identified with angiography and present a potential risk during embolization.17,18 JNA can have a high incidence of arteriovenous shunts and larger particles of PVA (’140 m in calibre) are recommend for preoperative embolization.5 Some authors argue for the need to perform direct intratumoral embolization in addition to preoperative embolization in cases of JNA with significant vascular contribution from the ICA.18 Temporary balloon occlusion of the ICA without neurologic complication was used in the preoperative management in 2 patients from our group. Little additional benefit was realized from direct intratumoral embolization with temporary balloon occlusion of ICA. Weighing in the potential complications and our brief experience, we do not believe this is necessary for the treatment of intracranial
585
JNA. A vascular supply from the ICA is not unusual for large tumors extending intracranially and does not limit complete and safe surgical excision.5,17,18 Comparable rates of complete removal of intracranial JNA to less aggressive disease were achieved. Radiologic follow-up is essential in the early identification of residual or recurrent disease.14,20 Postoperative MRI and contrast-enhanced CT have been recommended to reveal residual or recurrence tumor.21,22 Recurrence may also be demonstrated by carotid angiography or by subtraction MRI.11 In our institution, all patients have been studied with MRI ⫾ carotid angiography between 6 and 12 month postoperation. Subsequent follow-up is by CT or MRI yearly. Carotid angiography confirmed disease in 2 cases where MRI did not show recurrence. MRI, carotid angiography, and contrast-enhanced CT are all useful in the detection of residual or recurrent disease with little evidence to show the superiority of one modality.
CONCLUSION Preoperative imaging will often overstage advanced tumors. Advanced JNA can be successfully removed in 82% of patients during their first surgery despite intracranial extension. The majority of advanced JNA remain extradural, and a surgical plane can usually be found between important structures such as the ICA, cavernous sinus, and MCF. Where possible, an endoscopic approach maybe the first option for many patients. The endoscopic surgeon will also benefit from defined tumor-dural dissection planes. Anterior access, whether endoscopic or open, is often sufficient to address intracranial involvement. When an open anterior approach is used, the midface degloving technique affords excellent exposure even for advanced disease. Lateral approaches with their associated morbidity can be reserved for selected recurrent disease. Figures 2 to 4
AUTHOR INFORMATION From the Divisions of ORL and skull base surgery (Drs Danesi, Panciera, and Harvey) and Neuroradiology (Dr Agostinis), Ospedali Riuniti di Bergamo, Bergamo, Italy. Corresponding author: Richard Harvey, Flat 170, Green Place Flats, Dr Diogo de Faria, 1201, Vila Mariana, CEP 04037-004, Sao Paulo, SP, Brazil. E-mail address: [email protected].
AUTHOR CONTRIBUTIONS Giovanni Danesi, lead author, primary surgeon; Davide T. Panciera, surgical and clinical research participation; Richard J. Harvey, database, clinical research, statistics, manuscript preparation; Cristina Agostinis, radiologic expertise and interpretation.
586
Otolaryngology–Head and Neck Surgery, Vol 138, No 5, May 2008
FINANCIAL DISCLOSURE None.
REFERENCES 1. Bales C, Kotapka M, Loevner LA, et al. Craniofacial resection of advanced juvenile nasopharyngeal angiofibroma. Arch Otolaryngol Head Neck Surg 2002;128:1071–1078. 2. Radkowski D, McGill T, Healy GB, et al. Angiofibroma: changes in staging and treatment. Arch Otolaryngol Head Neck Surg 1996;122: 122–129. 3. Lee JT, Chen P, Safa A, et al. The role of radiation in the treatment of advanced juvenile angiofibroma. Laryngoscope 2002;112:1213–1220. 4. Marshall AH, Bradley PJ. Management dilemmas in the treatment and follow-up of advanced juvenile nasopharyngeal angiofibroma. ORL J Otorhinolaryngol Relat Spec 2006;68:273–278. 5. Schroth G, Haldemann AR, Mariani L, et al. Preoperative embolization of paragangliomas and angiofibromas: measurement of intratumoral arteriovenous shunts. Arch Otolaryngol Head Neck Surg 1996; 122:1320 –1325. 6. Andrews JC, Fisch U, Valavanis A, et al. The surgical management of extensive nasopharyngeal angiofibromas with the infratemporal fossa approach. Laryngoscope 1989;99:429 – 437. 7. Wormald PJ, Van Hasselt A. Endoscopic removal of juvenile angiofibromas. Otolaryngol Head Neck Surg 2003;129:684 – 691. 8. Danesi G, Panizza B, Mazzoni A, et al. Anterior approaches in juvenile nasopharyngeal angiofibromas with intracranial extension. Otolaryngol Head Neck Surg 2000;122:277–283. 9. Douglas R, Wormald P-J. Endoscopic surgery for juvenile nasopharyngeal angiofibroma: where are the limits? Curr 2006;14:1–5. 10. Close LG, Schaefer SD, Mickey BE, et al. Surgical management of nasopharyngeal angiofibroma involving the cavernous sinus. Arch Otolaryngol Head Neck Surg 1989;115:1091–1095.
11. Lloyd G, Howard D, Phelps P, et al. Juvenile angiofibroma: the lessons of 20 years of modern imaging. J Laryngol Otol 1999;113:127–134. 12. Donald PJ, Enepikedes D, Boggan J. Giant juvenile nasopharyngeal angiofibroma: management by skull-base surgery. Arch Otolaryngol Head Neck Surg 2004;130:882– 886. 13. Browne JD. The midfacial degloving procedure for nasal, sinus, and nasopharyngeal tumors. Otolaryngol Clin North Am 2001;34:1095– 1104. 14. Scholtz AW, Appenroth E, Kammen-Jolly K, et al. Juvenile nasopharyngeal angiofibroma: management and therapy. Laryngoscope 2001; 111:681– 687. 15. Fagan JJ, Snyderman CH, Carrau RL, et al. Nasopharyngeal angiofibromas: selecting a surgical approach. Head Neck 1997;19:391–399. 16. Laurent A, Wassef M, Chapot R, et al. Partition of calibrated tris-acryl gelatin microspheres in the arterial vasculature of embolized nasopharyngeal angiofibromas and paragangliomas. J Vasc Interv Radiol 2005; 16:507–513. 17. Petruson K, Rodriguez-Catarino M, Petruson B, et al. Juvenile nasopharyngeal angiofibroma: long-term results in preoperative embolized and non-embolized patients. Acta Otolaryngol (Stockh) 2002;122:96 – 100. 18. Lefkowitz M, Giannotta SL, Hieshima G, et al. Embolization of neurosurgical lesions involving the ophthalmic artery. Neurosurgery 1998;43:1298 –1303. 19. Casasco A, Houdart E, Biondi A, et al. Major complications of percutaneous embolization of skull-base tumors. AJNR Am J Neuroradiol 1999;20:179 –181. 20. Mann WJ, Jecker P, Amedee RG. Juvenile angiofibromas: changing surgical concept over the last 20 years. Laryngoscope 2004;114:291–3. 21. Howard DJ, Lloyd G, Lund V. Recurrence and its avoidance in juvenile angiofibroma. Laryngoscope 2001;111:1509 –1511. 22. Kania RE, Sauvaget E, Guichard J-P, et al. Early postoperative CT scanning for juvenile nasopharyngeal angiofibroma: detection of residual disease. AJNR Am J Neuroradiol 2005;26:82– 88.