University of Pittsburgh Medical Center (UPMC) staging system. Axial computed tomographic (CT) scans showing tumor confined to the left nasal cavity and medial pterygopalatine fossa (UPMC stage I) (A) and tumor involving left lateral pterygopalatine fossa (UPMC stage II) (B). Figure panels C and D are from the same patient in whom the axial CT scan (C) shows tumor involving the left intratemporal fossa and an angiogram (D) shows no residual vascularity after embolization of the external carotid artery tributaries (UPMC stage III). Figure panels E and F are from the same patient in whom the axial CT scan (E) shows tumor involving the infratemporal fossa and angiogram (F) with residual vascularity from the internal carotid artery after embolization of the external carotid artery tributaries (UPMC stage IV). Figure panel G is an axial CT scan showing tumor lateral to the cavernous internal carotid artery (UPMC stage VL).
Dot plots depicting correlation between the tumor stage using various staging systems and the total estimated intraoperative blood loss. A, University of Pittsburgh Medical Center stage10 (P < .001); B, Onerci stage5 (P = .04); C, Radkowski stage6 (P = .06); D, Andrews stage7 (P = .08); E, Chandler stage8 (P = .13); and F, Sessions stage9 (P = .07). Staged surgical procedures are counted as 1 data point. There are a total of 40 data points (some are superimposed) for 35 patients. Five patients had second surgical procedures for residual or recurrent disease. The correlation is determined using Spearman correlation.
A, Stage V angiofibroma with encasement of both internal carotid arteries and medial and lateral intracranial extension. B, At the first staged surgery, one-half of the extracranial tumor was excised. C, The remainder of the extracranial tumor was excised at the next stage, and residual intracranial tumor around the left internal carotid artery and lateral to the orbit was excised at the next stage. D, There is no evidence of residual or recurrent disease 2 years following surgery. A stable meningocele is noted at the site of tumor excision in the left middle cranial fossa. Residual enhancement of scar tissue adjacent to the internal carotid artery remains unchanged after 2 years.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Snyderman CH, Pant H, Carrau RL, Gardner P. A New Endoscopic Staging System for Angiofibromas. Arch Otolaryngol Head Neck Surg. 2010;136(6):588–594. doi:10.1001/archoto.2010.83
To develop a new staging system for juvenile nasopharyngeal angiofibroma that reflects changes in surgical approaches (endonasal), route of intracranial extension, and the extent of vascular supply from the internal carotid artery.
Retrospective review of case series.
Academic medical center.
Patients undergoing endoscopic endonasal surgery for juvenile nasopharyngeal angiofibroma at the University of Pittsburgh Medical Center (UPMC), Pittsburgh, Pennsylvania, from 1998 through 2008.
Patients were staged according to current systems and compared with a new staging system that also incorporated the route of skull base extension and tumor vascularity.
Main Outcome Measures
Estimated blood loss, number of operations, and tumor recurrence.
Skull base erosion was observed in 74% of cases. Following embolization of external carotid artery tributaries, residual vascularity from the internal carotid artery was seen in 51% of patients. Residual vascularity, classified as UPMC stage IV and V, strongly correlated with blood loss, requirement for multiple procedures, and residual or recurrent tumor.
Tumor size and extent of sinus disease are less important in predicting complete tumor removal with endonasal surgical techniques. The UPMC staging system for juvenile nasopharyngeal angiofibroma accounts for 2 important prognostic factors, route of cranial base extension, and vascularity and is applicable to endoscopic or open approaches. Compared with other staging systems, the UPMC staging system provides a better prediction of immediate morbidity (including blood loss and need for multiple operations) and tumor recurrence.
Juvenile nasopharyngeal angiofibromas are rare benign tumors that are almost exclusively found in male adolescents. They are characterized by high vascularity and locally aggressive growth with the potential for skull base erosion and intracranial extension. Common symptoms at presentation include nasal obstruction and epistaxis. Large tumors may also cause orbital displacement with diplopia, facial hypesthesia from trigeminal nerve compression, trismus, and facial deformity. Intracranial extension is seen in 10% to 20% of cases.1
The preferred treatment of juvenile nasopharyngeal angiofibroma is complete surgical excision, with reported recurrence rates of 15% to 50%.1,2 A variety of open surgical approaches have been described: transpalatal, midfacial degloving, LeFort I osteotomies, lateral rhinotomy, transfacial, infratemporal fossa, and craniofacial approaches. Endoscopic transnasal approaches have largely replaced open approaches for small tumors3 and may be combined with open approaches for larger tumors, including midfacial degloving approach with endoscopic excision. Treatment alternatives for large tumors that are deemed “inoperable” include radiation therapy and antiandrogen therapy.4
Multiple staging systems have been proposed for juvenile nasopharyngeal angiofibroma (Table 1).5-9 The common feature of these staging systems is that the tumor stage depends on the extent and sites of tumor involvement. Advanced stages are associated with skull base involvement and intracranial extension. Most of these staging systems were developed prior to the widespread adoption of endoscopic techniques, and thus the stages may not include important prognostic factors for endoscopic techniques. A major deficiency of existing staging systems is that they do not account for the route of intracranial extension and the vascularity of the tumor.
We performed a retrospective review of endoscopic surgical procedures for juvenile nasopharyngeal angiofibroma at our institution with the goals of evaluating prognostic factors for recurrence and morbidity and developing a new staging system that incorporates these factors and reflects changes in surgical management.
A retrospective review of surgical procedures performed for juvenile nasopharyngeal angiofibroma at the University of Pittsburgh Medical Center (UPMC) from 1998 through 2006 was previously performed10 to compare the outcomes of open, endoscopic-assisted, and completely endoscopic surgery. Embolization was performed 1 day before surgery using standard techniques with a combination of small particles for the tumor bed and coils for the proximal feeding vessels. Preembolization and postembolization angiograms were reviewed for evidence of residual vascularity from the internal carotid artery (ICA). Vascularity from branches of the external carotid artery was effectively abolished with the embolization procedure (bilateral embolization of internal maxillary arteries, facial arteries, and ascending pharyngeal arteries as indicated). The degree of residual vascularity from the ICA was not quantified but was based on identifiable vessels coming from the artery.
We included additional patients with juvenile nasopharyngeal angiofibroma who underwent endoscopic surgery from 2007 to 2008. Patients in both series without blood loss data were excluded. This database was then used to stage patients using existing staging systems and the UPMC staging system. The tumor stage was correlated with outcome variables including blood loss, need for multiple operations, and recurrence. In patients undergoing planned multiple operations (staged surgery), estimated total operative blood loss from each operation was pooled. Staged surgical procedures were not considered as residual or recurrent disease. Residual disease was defined as tumor remaining after definitive surgery and detectable on immediate postoperative scans and was initially managed with observation. Recurrent disease was defined as regrowth of tumor following a negative postoperative scan.
The UPMC staging system for juvenile nasopharyngeal angiofibroma is given in Table 2 and Figure 1. It acknowledges that paranasal sinus involvement is not an impediment to complete surgical excision. Stage I includes the smallest tumors with no significant extension beyond the site of origin and remain medial to the midpoint of the pterygopalatine space. Stage II tumors extend to the paranasal sinuses and lateral to the midpoint of the pterygopalatine space. Together, stages I and II can be considered minimal stage tumors. For this staging system, it is presumed that all but the smallest of tumors will have preoperative embolization of the internal maxillary artery and other contributing branches of the external carotid arterial system to devascularize the tumor and facilitate surgery. Tumors that are locally advanced with skull base erosion or extension to additional extracranial spaces, including orbit and infratemporal fossa, but no residual vascularity following embolization are stage III. Technically, the surgery required for a stage III tumor requires greater access and is more challenging but is not plagued by excessive bleeding. The tumor usually has a well-defined margin that easily separates from dura, periorbita, or muscle fascia. Stage IV and V tumors are characterized by residual vascularity from the intracranial circulation following embolization. True intracranial extension (stage V) is subdivided into medial (medial cavernous sinus) and lateral (middle fossa) routes of extension, since this has significant implications for surgical access, potential morbidity, and risk of residual disease. Medial extension is defined by tumor extension medial to the paraclival and cavernous segments of the ICA and can include extension into the cavernous sinus posterior to the ICA at the level of the pituitary gland. Lateral extension is defined by tumor extension to the middle cranial fossa superior to the petrous ICA and lateral to the paraclival and cavernous segments of the ICA. It may occur by direct extension with erosion of the floor of the middle cranial fossa or by extension through foramina (eg, foramen ovale, superior orbital fissure).
Outcome variables and tumor stage were correlated using the Spearman correlation, with a significance level of .05 (GraphPad Prism software, version 4.0a for Macintosh; GraphPad Software Inc, La Jolla, California). Multiple operations and tumor stage were compared with a Fisher exact test.
Thirty-five patients were included in this study and their clinical characteristics and UPMC tumor stage are summarized in Table 3 and Table 4. Skull base erosion was observed in 74% of cases. Embolization of the external carotid artery tributaries was successful in all cases. Thus, following embolization, residual vascularity from identifiable tributaries of the ICA was seen in approximately 51% of patients.
The UPMC stage had the strongest correlation with intraoperative blood loss (P < .001) (Figure 2). In contrast to other staging systems, all patients with intraoperative blood loss greater then 1000 mL were only in the most advanced stages (IV and V).
Planned multiple operations were necessary in 9 of 35 patients (26%) to achieve complete tumor removal (Table 3). The decision to stage an operation was based on the residual vascularity of the tumor following embolization and intraoperative blood loss. There was a good correlation between UPMC stage and the need for multiple operations (P < .001) (Table 5).
Residual disease was detected in 5 of 35 patients (14%) following surgery (Table 3). In 3 of these patients, the residual tumor remained stable on serial scans and further therapy was not necessary; 2 patients had additional surgery to achieve excision. Delayed recurrence was observed in 3 patients and all of these patients had additional surgery (Table 3). The location of residual and recurrent tumor was in the intracranial component of the tumor in all cases. Five patients had tumor medial to the ICA, and 3 patients had tumor lateral to the ICA but in close proximity to cranial nerves. The need for multiple operations, whether staged, residual, or recurrent tumor, was highly correlated with UPMC stage (16 of 17 patients) (Table 4). Compared with other staging systems, the UPMC stage had the highest correlation with recurrent and residual tumor (7 of 8 patients with stages IV and V) (Table 6).
The goals of treatment of juvenile nasopharyngeal angiofibroma are complete surgical excision with minimal morbidity and avoidance of radiation therapy. The choice of a surgical approach is designed to provide access to routes of tumor extension and control of the vascular supply of the tumor. Secondary goals of the surgical approach include avoidance of morbidity including blood loss and transfusion, cranial nerve injury, disruption of growth centers, cosmetic deformity, brain manipulation, and soft-tissue injury with associated pain. Successful surgery can obviate any consideration of radiation therapy with its attendant adverse effects and long-term risk of radiation-induced neoplasia in a young patient population.
The primary purpose of a tumor staging system is to provide prognostic information for treatment planning and counseling of patients. Secondary objectives include prediction of outcome (tumor recurrence) and promotion of uniform reporting in the medical literature to allow comparison of patient series with different tumor sites and treatments.
The UPMC staging system incorporates the following important prognostic factors that were not addressed by prior staging systems: route of extension and residual vascularity following embolization. The route of extension, including medial or lateral routes relative to the vertical segments of the ICA, is an important consideration in selecting the optimal surgical approach. For example, a tumor that extends medial to the ICA with invasion of the sella or cavernous sinus requires a midline approach. Tumors that extend laterally to the middle cranial fossa may be approached using an endonasal or an infratemporal approach.
The most important outcome in the operative period is blood loss and is the primary determinant of planned multiple surgical procedures. Operative deaths from hemorrhage undoubtedly occur and could perhaps be prevented with better prognostication and selection of alternate treatment strategies or referral to more experienced centers. A comparison of existing staging systems demonstrates no consistent relationship between tumor stage and the estimated blood loss. The morbidity of surgery is dependent on the stage of the tumor and the surgical approach. Observed morbidities in this patient population included trismus, infraorbital nerve hypesthesia, palatal hypesthesia, and epistaxis. No effects on facial growth were observed. We were unable to correlate these additional operative morbidities with tumor stage owing to the small number of patients with deficits and lack of consistent documentation.
The most important outcome in the postoperative period is tumor persistence or recurrence and may be a consequence of tumor location as well as vascularity. Almost all of the recurrences in our series occurred in patients with skull base involvement and residual vascularity. Dissection in proximity of the ICA when there is significant residual vascularity is more difficult owing to impaired visualization and surgeons are more likely to leave residual tumor that may be apparent on immediate postoperative scans or result in delayed recurrence. Compared with other staging systems, the UPMC staging system provided superior prediction of recurrent and residual disease.
Although the UPMC staging system was designed based on patients who underwent predominantly endoscopic resection, it is not influenced nor limited by the type of surgical approach. Regardless of approach used, residual vascularity and location of tumor medial to the ICA are key issues that limit surgery to avoid complications and significant morbidity. According to the Radkowski staging system, 74% of the tumors in this series would be stage III (very advanced tumors) owing to presence of skull base erosion. However, taking vascularity into account, 51% would be considered advanced using the UPMC staging system (effectively down-staging the tumor). Thus, not all cases with skull base erosion had residual vascularity and consequently, were amenable to endoscopic resection in a single operation.
Surgical strategies for dealing with advanced-stage angiofibromas with skull base erosion and residual vascularity include preoperative embolization, sequential dissection of tumor segments, and staging of surgery. Tumor segments are defined by vascular territories and each territory is addressed sequentially. For example, surgical excision of a large angiofibroma with bilateral contribution from the ICA would consist of creation of a midline corridor with identification of skull base landmarks, unilateral excision of extracranial tumor, contralateral excision of extracranial tumor, unilateral excision of intracranial tumor, and contralateral excision of intracranial tumor (Figure 3). Staging of surgery could be performed at any step depending on intraoperative blood loss. The UPMC staging system is useful in developing a surgical strategy for individual patients.
In conclusion, the UPMC staging system incorporates important prognostic information for the surgeon such as the residual vascularity of the tumor following embolization and route of skull base extension. Blood loss and need for transfusion is strongly correlated with UPMC stage. Staged operations were only necessary in tumors with residual vascularity, and recurrent residual disease was only observed in advanced stages (III-V). The UPMC staging system may be superior to other staging systems in predicting multiple prognostic factors of patients undergoing endoscopic surgery for juvenile nasopharyngeal angiofibroma.
Correspondence: Carl H. Snyderman, MD, Department of Otolaryngology, Eye and Ear Institute, 200 Lothrop St, Ste 500, Pittsburgh, PA 15213 (email@example.com).
Submitted for Publication: August 19, 2009; final revision received December 31, 2009; accepted January 27, 2010.
Author Contributions: Dr Snyderman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Snyderman, Carrau, and Gardner. Acquisition of data: Snyderman, Pant, and Gardner. Analysis and interpretation of data: Snyderman, Carrau, and Gardner. Drafting of the manuscript: Snyderman, Pant, and Carrau. Critical revision of the manuscript for important intellectual content: Snyderman, Carrau, and Gardner. Administrative, technical, and material support: Carrau. Study supervision: Snyderman.
Financial Disclosure: None reported.