Juvenile nasopharyngeal angiofibroma samples (from case 10) under immunohistochemical staining for CD31 (A), Ki67 (B), and vascular endothelial growth factor (VEGF) (C) and their corresponding positive control stainings: D, under CD31 stain, the normal kidney glomeruli and microvessels have been stained; E, under Ki67 stain, squamous cell carcinoma proliferating cells, predominantly in tumor islets, are stained; and F, under VEGF stain, normal kidney cells in the glomeruli and vessel endothelium are stained. Original magnification ×100 for all panels.
Brieger J, Wierzbicka M, Sokolov M, Roth Y, Szyfter W, Mann WJ. Vessel Density, Proliferation, and Immunolocalization of Vascular Endothelial Growth Factor in Juvenile Nasopharyngeal Angiofibromas. Arch Otolaryngol Head Neck Surg. 2004;130(6):727-731. doi:10.1001/archotol.130.6.727
Juvenile nasopharyngeal angiofibroma (JNA) is a rare, highly vascularized neoplasm of the nasopharynx that affects boys and young men. The underlying dysregulated molecular mechanisms remain unclear. The participation of angiogenic growth factors has been suggested, but few studies have been published.
To evaluate the expression and localization of vascular endothelial growth factor (VEGF), proliferating cells, and vessel density in JNA.
Immunohistochemical examination of 10 consecutive JNAs (8 primary tumors and 2 recurrent tumors).
Paraffin-embedded and cryopreserved JNA samples were included. VEGF-, CD31-, and Ki67-specific antibodies were applied and visualized using light microscopy. Vascularization was determined by counting CD31-positive vessels. Proliferating and VEGF-expressing vessels as well as stromal cells were quantified by the same method. Patients' age at the time of surgery and tumor stage were correlated with the immunohistochemical data.
All tumors were heavily vascularized, but major differences were noted between the samples. About half of the vessels were proliferating (Ki67 positive) and half of the Ki67-positive cells were also VEGF positive. The tumor stroma was VEGF positive in 8 of 10 samples and proliferating in 5 of these 8. The 5 samples with both VEGF- and Ki67-positive stroma showed high vessel densities. No correlation was observed between age or tumor stage and vessel density, VEGF expression, or Ki67 expression.
In JNA, VEGF is frequently expressed by stromal cells and vessels and is associated with proliferation and increased vessel density. We suggest the promotion of vascularization by VEGF, but the involvement of androgens in JNA angiogenesis still needs to be analyzed.
Juvenile nasopharyngeal angiofibroma (JNA) is a rare, histologically benign neoplasm of the nasopharynx that affects mostly boys and young men. A common feature of these tumors is their strong vascularity. However, the pathogenesis of JNA remains unclear. A role for androgens in JNA has been proposed but not proven.1- 5 Another influence might be the expression of angiogenic growth factors by the tumor that leads to vessel growth and subsequent tumor proliferation. This hypothesis is supported by a study conducted by Schiff and colleagues6 in 1992 that documented immunolocalization of basic fibroblast growth factor, a strong proangiogenic cytokine, in the vessel endothelium of JNA. In another study performed by Nagai and colleagues7 using a polymerase chain reaction approach, the messages for several proangiogenic cytokines were detected, but the expressing cells were not defined because of inherent limitations of the technique used. Recently, Dillard and colleagues8 reported the expression of transforming growth factor β1 in JNA stroma and endothelial cells. However, these studies do not address the type and localization of growth factor–expressing cells and the potential correlation with vessel density and tumor proliferation. Moreover, the most prominent proangiogenic growth factor in tumor biology—vascular endothelial growth factor (VEGF)—has not as yet been analyzed.
The growth of solid tumors depends on vessel growth as a prerequisite of tumor cell proliferation.9 Vascular endothelial growth factor is frequently expressed by tumor cells, including glioblastoma,10 colon carcinoma,11 renal cell carcinoma,12 breast carcinoma,13 paraganglioma (unpublished data, 2003),14 as well as by many normal tissues, including gastrointestinal smooth muscle, lung, alveolar epithelium, stomach, renal collecting tubule, colon mucous epithelium, and cardiac myocytes.15
In the present study, we analyzed the vessel density and the expression and localization of VEGF and the proliferation marker Ki67 in JNA. We found high vessel densities along with strong proliferating and VEGF-expressing vessel endothelium cells associated with VEGF-expressing and proliferating stromal cells.
The study included 10 samples of 9 consecutive patients who underwent surgery between August 1999 and April 2003 in Mainz, Germany; Poznan, Poland; and Holon, Israel. The age of the patients at the time of surgery ranged from 15 to 33 years (mean age, 21 years). One patient underwent surgery again after 12 months because of a relapse; this patient's primary and recurrent JNAs were labeled case 3 and case 10, respectively. Case 4 was also a recurrence, but no sample of the primary tumor was available for analysis. Tumor stages were classified according to Fisch16 (Table 1).
Tissues were either snap frozen in liquid nitrogen immediately after resection and stored until further use at –80°C or paraffin embedded after fixation. For immunohistochemical staining, cryopreserved samples were sectioned, equilibrated to room temperature, fixed in acetone at −20°C for 10 minutes, air dried, and washed in 0.05% polysorbate (Tween) solution with Tris-buffered saline (TBS-Tween) for 5 minutes. Paraffin-embedded samples, after dewaxing and rehydration, were treated by microwave for antigen retrieval (three 5-minute cycles at 600 W in 10mM citrate buffer, pH 6.0). Endogenous peroxidase was inhibited by immersing slides in 3% hydrogen peroxide methanol solution for 20 minutes. Slides were washed for 5 minutes in distilled water and for 5 minutes in TBS-Tween.
After preincubation with 10% normal serum in 1% bovine albumin phosphate-buffered saline solution for 1 hour to avoid unspecific binding, the primary antibodies (Table 2) were stored overnight at 4°C. Dilutions of antibodies were prepared in 1% bovine albumin phosphate-buffered saline solution at room temperature. Slides were washed twice with TBS-Tween and consecutively incubated with biotinylated secondary antibody for 30 minutes; again washed twice with TBS-Tween and incubated with streptavidin horseradish peroxidase conjugate (DAKO, Hamburg, Germany) for 30 minutes; washed twice again with TBS-Tween and finally incubated with 1.85mM diamino benzidine/hydrogen peroxide (Sigma, St Louis, Mo) for 1 minute.
Immediately after the staining developed, slides were washed with distilled water for 5 minutes and counterstained with hematoxylin (1:5 in phosphate-buffered saline; Merck, Darmstadt, Germany). Slides were rinsed for 5 minutes with distilled water and dehydrated for 3 to 5 minutes each with 80% → 100% isopropanol. Finally, samples were immersed twice in xylol (5 minutes each time), closed with a coverslip, and embedded with Enthelan (Merck).
Vascular endothelial growth factor stainings were analyzed and documented using an inverted microscope (Zeiss, Jena, Germany), and images were saved as jpg files. For microvessel density, the vessels marked by CD31 were counted in 3 "hot spot" fields of view (areas of high vessel density) at ×100 magnification (1 mm2), and the mean counts were calculated. Ki67-positive vessels and VEGF-positive vessels were counted by the same method. Expression of VEGF and Ki67 by tumor stroma cells was graded as positive or negative at ×400 magnification. Sections of normal kidney and squamous cell carcinoma served as positive controls. Sections incubated without the primary antibody served as negative controls (data not shown). Countings were performed by the same person to achieve minimum variability.
We analyzed 10 samples, corresponding to 9 patients, including 2 samples from recurrent tumors. Vessel densities were quantified by CD31 staining. Localization and quantification of proliferating or VEGF-expressing vessels and stromal cells were determined by Ki67 and VEGF staining (Figure 1). Patient age at the time of surgery and tumor stage were correlated with each immunohistochemical result (Table 1).
Patient ages ranged from 15 to 33 years (mean age, 21 years). Most tumors were classified as stage II or III, with only 1 sample each at stages I and IV. No correlation was observed between age or tumor stage and vessel density, VEGF expression, or Ki67 expression. Vessel density and numbers of proliferating and VEGF-expressing vessels varied widely among the samples (factors 4 to 20): Vessel counts defined by CD31 staining varied from 16 per field of view to 99 (median count, 55); the number of proliferating vessels ranged from 12 to 48 per field (median count, 23); and the VEGF-expressing vessel counts ranged from 2 to 39 per field (median count, 13). In other words, about 50% of the tumor vessels were proliferating (Ki67 positive), and of these about 50% were VEGF positive. The tumor stroma showed a VEGF-positive staining in most samples, with the exceptions of samples 7 and 8 and a staining of single cells in sample 2. Strong proliferative activity in stromal cells was observed in 3 cases (samples 3, 5, and 9), and in single cells of samples 1 and 2.
The 5 samples that showed stromal VEGF expression and proliferation showed high vessel densities (43-99 vessels per field). The samples without VEGF and/or Ki67 stromal stainings conversely showed low vessel densities (16-32 vessels per field), with 1 exception: sample 7 had high vessel densities (52 vessels per field) despite negative stromal stainings for CD31 and Ki67. Similar observations were made for Ki67- and VEGF-positive vessel counts: those samples with stromal VEGF staining and Ki67 staining showed higher counts for proliferating (20-42 vessels per field) and VEGF-expressing (12-39 vessels per field) vessels than the samples that were negative for either Ki67 alone or for both Ki67 and VEGF stainings. Again, 1 exception was noted: case 2 showed low Ki67 and VEGF vessel counts (13 and 5, respectively) despite stromal staining. On the other hand, the previously aberrant sample, case 7, showed low counts for Ki67- and VEGF-positive vessels, in accordance with negative stromal staining. Interesteringly, this sample was only positive for VEGF and Ki67 with single stromal cells, which might explain the low tissue VEGF levels.
In the vicinity of the largest vessels (50-200 µm), it was rare for additional vessels to be observed. Often the larger vessels were observed in the center of the sample, while smaller vessels were more peripheral (data not shown).
In this series of JNA specimens, we observed high vessel densities associated with stromal VEGF and Ki67 expression. The participation of androgens and angiogenic growth factors has been suggested in the growth of JNA because of the sex-dependent occurrence and the high vessel densities observed. However, the results of antiandrogen therapies and of immunohistochemical analyses of androgen receptors remain conflicting. One study reported the regression of JNA after antiandrogen therapy in 4 of 5 patients.3 Another comprehensive study demonstrated the expression of androgen receptors in most of the samples in the vessel endothelium as well as in the tumor stroma cells, which might indicate a role for these receptors in the pathophysiology of JNA.5 In contrast, Gatalica4 compared androgen receptors in JNA samples with normal turbinate tissue and found similar immunoreactivity in both groups in stromal and endothelial cells.
A mechanism of androgen action might be the induction of angiogenic growth factors that leads to the observed high vessel densities in JNA. Lissbrant and colleagues17 reported strong endothelial proliferation in rat reproductive organs during testosterone treatment and decreased proliferative activity after testosterone treatment was stopped. In a recent ex vivo study conducted on lung vascular endothelial cells,18 the induction of endothelial cell proliferation by testosterone could be demonstrated in endothelial cells isolated only from male rats. Female rats were completely unresponsive to testosterone, which indicates a sex-dependent responsiveness of endothelial cells to androgen stimulation. In another study, Häggström and colleagues19 reported that testosterone stimulates rat prostate endothelial cell proliferation and vascular growth and that this effect is probably mediated through the induction of VEGF synthesis by testosterone.
The secretion of proangiogenic growth factors by tumors is well established and in fact is a prerequisite for progression and metastasis in many tumors.20- 22 However, several tumor types—frequently benign ones—secrete proangiogenic factors, which leads to vessel growth but has little or no impact on tumor growth.23,24 For the benign but strongly vascularized JNA, limited data concerning growth-factor secretion are available.6,8 We observed strong expression of VEGF predominantly in the vessel endothelium of JNA and lower levels in the tumor stroma. This observation is in accordance with the findings of Schiff,6 who described the expression of basic fibroblast growth factor, the second most prominent vessel growth factor, in the endothelium. Dillard and colleagues8 reported the expression of transforming growth factor β1, another growth factor with proangiogenic and proliferation-promoting activity in both JNA tumor stroma and in vessel endothelium. We found a high degree of congruence of VEGF-expressing and proliferating endothelial cells. Interestingly, the tumors with high vessel densities (and high proportions of proliferating vessels) showed VEGF-expressing and proliferating stromal cells. This finding might be indicative of a proliferation-promoting activity of the tumor stroma cells. One might speculate that stromal VEGF leads to the stimulation of endothelial cell growth and VEGF expression by an autocrine loop as has been reported in human placenta vessels.25
We found no correlation between tumor stage and vessel density. This is in line with observations in other benign tumors where a correlation of proangiogenic factors with angiogenesis, but not with tumor size, has been shown (unpublished data, 2003).23,24 Therefore, in JNA the expression of VEGF might also lead to high vessel densities but not necessarily to large or aggressive tumors.
Another frequent observation was that of vessel-free spaces surrounding large, mostly round vessels. The average distance from large vessel to large vessel was 100 to 200 µm. That is equal to the distance that 2 vessels are able to support their interadjacent tissues with nutrients and oxygen.26 Conversely, this might suggest that the small, densely arranged, and irregularly shaped vessels might have restricted functionality, maybe because of dysregulated vessel growth as a consequence of overproduction of proangiogenic growth factors.
Collectively, the published animal data concerning the sex hormone dependence of endothelial cell growth, together with our findings and the data of others, support the view that JNA vascularization might be promoted by the action of angiogenic growth factors secreted by stromal cells, perhaps via the stimulation of androgens like testosterone. An autocrine loop of vessel endothelium expression of epidermal growth factor and therefore growth might be feasible. Other angiogenic factors like the reported basic fibroblast growth factor and transforming growth factor β16,8 should also be considered, given our observation that 2 of 10 tumors (cases 2 and 7; Table 1) showed high vessel densities despite low Ki67 and VEGF vessel counts and negative or rare Ki67 and VEGF stromal staining.
The results of the present study increase our understanding of the pathophysiology of JNA. We conclude that VEGF is secreted by JNA and suggest that VEGF contributes to the strong vascularization of this benign tumor. Therefore, antiangiogenic therapy might be considered. However, the impact and action of androgens on JNA vascularization requires analysis.
Corresponding author and reprints: Jürgen Brieger, PhD, Department of Otorhinolaryngology, Laboratory of Molecular Tumor Biology, University Hospital of Mainz, Germany, Langenbeckstrasse. 1, 55101 Mainz, Germany (e-mail: email@example.com).
Submitted for publication August 22, 2003; final revision received November 3, 2003; accepted November 5, 2003.
This study was supported by a grant from the Head and Neck Tumor Research Foundation, Wiesbaden, Germany.
We thank Karin Bakes for technical assistance and Lisa A. Orloff, MD, for critical reading of the manuscript.