Blood Vessel Maturation in Human Uveal Melanoma: Spatial Distribution of Neovessels and Mature Vasculature

Abstract

Purpose

The aims of this study are (1) to evaluate the spatial distribution of neovessels and mature vessels in human uveal melanoma tumors and (2) to determine whether vessel maturation is associated with the major indicators for poor prognosis.

Methods

Immunohistochemical analyses were performed on human tissue specimens from enucleated eyes (n = 14) to assess total vessels, neovessels, mature vessels, and cell proliferation. Tumor morphology was analyzed by hematoxylin and eosin and modified periodic acid-Schiff (PAS) staining. The spatial distribution of neovessels and mature vessels was analyzed by immunohistochemistry, and correlated with major indicators of poor prognosis (i.e., aggressive PAS patterns, epithelioid cytology, mitotic figures, extraocular extension, anterior tumor location, ciliary body involvement, large tumor size, cell proliferation, and angiogenic activity).

Results

Neovesseldensity was greater than mature vessel density in apical (p = 0.17), central (p = 0.036), and peripheral (p = 0.31) regions of the tumors, while mature vessel density was greater than neovessel density in basal areas of the tumor (p = 0.47). This pattern indicated that vessel maturation begins at the base of the tumor and later extends to the peripheral and apical regions. The difference between mature and neovessel densities for the apical (−0.8 ± 1.9) and central areas (−0.8 ± 1.3) of the tumor was significantly higher than the difference obtained for the basal area (0.3 ± 1.6; p = 0.014 and p = 0.012, respectively), indicating a higher density of mature vessels compared to neovessels at the base. Statistical correlations were found between mature vessel density and tumor size (r = 0.48, p = 0.084), cell proliferation (r = 0.62, p = 0.042), and mitotic figures (r = 0.76, p = 0.001).

Conclusions

Significant differences exist in the spatial distribution of mature versus neovessels in human uveal melanoma. Vessel maturation is associated with known clinical and pathologic indicators of poor prognosis (e.g., cell proliferation). Antiangiogenic therapy should be considered for the treatment of ocular malignancies; however, the results of this study indicate that blood vessel maturation heterogeneity may limit the efficacy of vessel targeting agents.

Introduction

Uveal melanoma occurs in approximately 1,200 patients per year in the United States. Although rare, this is the most common primary intraocular malignancy in adults [1, 2], with large tumors carrying a mortality rate of almost 50% due to metastasis [3,4,5]. Efforts have been made to develop therapeutic tools to treat the tumor more effectively. Currently, 30–50% of cases with uveal melanoma undergo enucleation due to uncontrolled tumor proliferation.

A number of factors have been associated with poor prognosis and death due to metastasis in uveal melanoma. These include aggressive periodic acid-Schiff (PAS) patterns [6], epithelioid cytology [4], mitotic figures [7, 8], tumor size [9, 10], extraocular extension [4, 11], anterior tumor location [12,13,14], ciliary body involvement [4, 15], macrophage density [16], high microvascular density [17], cell proliferation [18], angiogenic activity [18, 19], and chromosomal abnormalities (e.g., monosomy 3) [20]. Among these factors angiogenic activity is of interest since novel therapeutic agents currently under study include vessel targeting agents.

Angiogenesis, the formation of new blood vessels from preexisting ones, has been widely recognized to be a biological indicator of tumor formation and proliferation [18,21,22,23]. Since angiogenesis plays an essential role in tumor proliferation and metastasis, attention has been focused on antiangiogenic treatments in a number of ocular tumors [24,25,26,27,28]. However, reports from studies in a mouse model of pancreatic islet cancer suggest that antiangiogenic therapy mainly prevents tumor growth with less effect on tumor regression [29]. In a mouse model of retinoblastoma, we have shown that the antiangiogenic steroid derivative anecortave acetate effectively reduces tumor burden in small tumors with high angiogenic activity [30]. However, this effect was diminished in larger tumors with significant tumor blood vessel maturation [31]. In other models, adding pericyte inhibitors to target mature vasculature in addition to antiangiogenic agents enhances tumor regression [29]. In an orthotopic model of uveal melanoma, selective targeting of pericytes reduces tumor vessel density and tumor burden [32].

The vascular system of uveal melanomas is highly complex including blood vessels and tumor-lined channels. Vessel density has been detected in these tumors using markers against CD34 [17] and factor VIII-related antigen [33]. Folberg et al. [34] and Maniotis et al. [35] demonstrated that aggressive melanoma cells create tumor-lined channels which are similar to blood vessels, but lack endothelial cells (ECs). The formation of these tumor cell-lined, PAS-positive channels is called ‘vasculogenic mimicry’. Recently, angiogenic activity was assessed by the detection of neovessels using CD105 immunoreactivity [18, 19]. Angiogenic activity in these tumors has been significantly associated with death due to metastasis [18, 19]. Mature vessels have been also detected in these tumors by staining with α-smooth muscle actin (α-SMA) [17]. The presence and spatial distribution of neovessels as well as mature vessels in human uveal melanoma tumors may impact the efficacy of antiangiogenic therapies, and may dictate the therapeutic strategies used. To the best of our knowledge, the spatial distribution of vessel maturation in uveal melanoma has not been assessed.

The aims of this study are to (1) evaluate the spatial distribution of neovessels versus mature vessels inthis malignancy and (2) determine if vessel maturation is associated with the major indicators for poor prognosis in uveal melanoma.

Methods

Human Tissue Specimens

This study was conductedin adherence with the tenets of the Declaration of Helsinki and the University of Miami Institutional Research Board. Human uveal melanoma tissue samples were consecutively obtained between the dates September 21, 2005 and August 27, 2006, from 14 patients (7 females and 7 males; median age: 56.5 years; age range: 13–86 years) who underwent enucleation due to extensive tumor at presentation at the Bascom Palmer Eye Institute (table 1).

Table 1.

Summary of tumor characteristics

Tumor No.	Tumor area mm2	Anatomic location	Histologic typea	Mitotic figures per 40 HPF	Extraocular extension, % scleral involvement	Location	Average grading densities			Aggressive vascular patterns loops/networks
							α-SMA	CD105	Ki-67
1	45.9	C, CB	mixed	11		base	5.0	4.3	3.8
						center	3.7	3.7	3.8
						apex	5.0	3.7	4.1
						periphery	5.0	4.7	3.9
2	28.8	C	spindle	0	50b	base	1.0	1.7	2.7
						center	1.0	1.7	2.9	+
						apex	2.0	1.7	2.2
						periphery	1.8	1.3	2.5
3	132.9	C, CB	mixed	4	15	base	3.0	3.0	3.5	+
						center	3.0	2.3	3.4
						apex	3.0	3.3	3.0
						periphery	3.5	2.2	3.1	+
4	32.4	C, CB	mixed	2	10	base	2.3	2.0
						center	2.0	3.0
						apex	1.0	3.0
						periphery	3.0	2.7
5	50.4	C	mixed	2	50	base	3.7	4.0	1.9	+
						center	1.3	5.0	1.2	+
						apex	1.0	5.0	1.2
						periphery	2.0	4.8	1.9
6	54.2	C, CB	mixed	3	<10	base	5.0	3.0	2.0
						center	2.0	3.7	1.0
						apex	2.7	3.7	2.0
						periphery	3.0	3.0	1.5
7	49.1	C	mixed	3	60	base	2.0	3.0
						center	2.0	4.0
						apex	1.0	4.7
						periphery	2.7	3.7
8	70.9	C, CB	mixed	2	100	base	5.0	4.3	2.4
						center	3.3	4.0	2.9
						apex	3.0	4.0	3.0
						periphery	3.5	4.0	2.7
9	53.8	C, CB	mixed	3	100	base	1.7	3.7	1.9
						center	2.7	3.0	1.5
						apex	2.3	2.7	1.6
						periphery	4.3	3.3	3.2	+
10	59.8	C	mixed	5	10	base	3.0	2.0
						center	2.0	2.3
						apex	3.0	2.0		+
						periphery	2.5	2.8
11	5.3	CB	spindle	0	60	base	5.0	3.7	3.0
						center	3.7	4.3	2.0	+
						apex	1.0	4.0	2.0	+
						periphery	3.0	3.8	1.0	+
12	18.4	C	mixed	1	10	base	1.0	3.3	1.0
						center	2.0	3.7	1.0
						apex	2.7	4.3	4.0
						periphery	1.8	4.0	1.5
13	23.4	C	spindle	6	10	base	4.7	1.0	2.0
						center	3.3	1.7	2.0
						apex	3.7	2.0	3.0
						periphery	3.0	3.3	3.0	+
14	302.6	C	mixed	7	100	base	5.0	4.0	4.0	+
						center	3.3	4.0	1.0
						apex	5.0	3.0	2.0	+
						periphery	4.0	4.0	3.0

C = Choroidal; CB = ciliary body.

^aHistologic types included spindle cells, epithelioid or mixed (spindle plus epithelioid).

^bExtension of tumor also noted through a vortex vein.

The following factors were considered indicators of poor prognosis and were correlated with the presence and distribution of mature tumor vessels: highly aggressive PAS patterns (networks and loops) [6], epithelioid cytology [4], mitotic figures [7, 8], extraocular extension [4, 11], anterior tumor location [12,13,14], ciliary body involvement [4, 15], large tumor size [9, 10], cell proliferation [18], and angiogenic activity [18, 19].

Hematoxylin and Eosin and Modified PAS

Enucleated eye specimens were fixed with 10% formalin, embedded in paraffin, and sectioned (8 μm). Slides were stained with hematoxylin and eosin (H&E) and modified PAS. The H&E stain was performed in previously bleached tissues and the PAS stain was modified by omitting the Harris hematoxylin or light green counterstain [34].

Each tumor area (i.e., base, center, apex, peripheral edges) was examined using a green filter for the distribution of the following PAS patterns: silent, straight, parallel, parallel with cross-links, arcs, arcs with branching, loops, and networks. These patterns have been previously described [6]. Although there is some debate regarding which PAS patterns are associated with worse prognosis, loops and networks have been shown by various groups to correlate significantly with metastatic death [4,5,6, 36, 37]. Thus, for simplicity of analysis, we divided PAS-positive patterns into highly aggressive (i.e., loops and networks) and less aggressive (i.e., straight, parallel, parallel with cross-links, arcs, and arcs with branching) categories. Images were digitally acquired at 40× and 200× magnifications (Olympus American, Melville, N.Y., USA).

Tumor morphology was analyzed by H&E staining. In order to measure tumor area, light-microscopic examination was performed on all histopathologic sections in a masked fashion. Images of all H&E- and PAS-stained sections were obtained with a digital camera at a magnification of 40×. Tumor boundaries were traced from H&E-stained sections and areas analyzed using Image Pro Express Software (Media Cybernetics, Silver Spring, Md., USA).

Immunohistochemistry

In addition to paraffin-embedded tissue, a portion of each tumor in which the architecture was preserved was frozen in OCT by the pathologist immediately after enucleation and serially sectioned (8 μm). Slides were fixed with methanol for 10 min (−20°C) and immunohistochemical analyses were performed.

Total vessels were detected with Alexa Fluor 568-conjugated lectin (Bandeira simplicifolia, a panendothelial binding agent; 1:1,000; Invitrogen, Carlsbad, Calif., USA) [38] and anti-collagen type IV (1:3,000; Sigma Chemical Co., St. Louis, Mo., USA). Mature vessels were detected with α-SMA Cy3 conjugate (1:3,000; Sigma Chemical Co.) which specifically binds to pericytes [39]. Neovessels were detected with anti-CD105 (clone SN6, 1:500; Abcam, Cambridge, Mass., USA) which has been shown to have specificity for EC displaying high angiogenic activity in uveal melanoma when compared to other markers (i.e., CD34) [18, 40]. Proliferating cells were labeled with Ki-67 (1:300; Abcam). Alexa Fluor 568 goat anti-mouse, 488 donkey anti-mouse, and 546 goat anti-rabbit were used as secondary antibodies for anti-collagen type IV, CD105, and Ki-67, respectively (1:500; Invitrogen). Omission of the primary antibody (secondary only) was used as a negative control for nonspecific binding. Cell nuclei were stained for 5 min with 4′,6′-diamidino-2-phenylindole (DAPI, 1:5,000; Invitrogen).

Scanning of Immunofluorescence and Image Processing

Serial cross sections of the tumors were examined for the presence of the different markers with a BX51 Olympus upright fluorescence microscope (Olympus American). Tumor cross sections were sequentially scanned by moving the stage in a stepwise fashion to take overlapping pictures of the entire tumor section. Each field was photographed 3 times at either 40× or 200× magnification using different filters for the DAPI, Alexa Fluor 488, and 568 signals. Montage images of the tumor, which consisted of 36 or 64 fields (6 × 6 or 8 × 8 tiled images, depending on the size of the tissue section), were digitally created with Image-Pro Discovery software.

Tumor Vessel Density Grading

Three photographs taken at 200× magnification, with an area of 0.137 mm² (0.428 × 0.320 mm), were analyzed per each tumor area. The basal, central, apical, and peripheral areas of each tumor were chosen to analyze the vessel densities and spatial configuration of neovessels and mature vessels. Montages of the total tumor were created and the tumor was divided into the five specified areas as depicted in figure 1A. Tumor neovessel and mature vessel density was graded in a masked fashion in the designated regions of the tumors by three investigators (Y.P., C.M.C., A.A.). Neovessels and mature vessel densities were graded using a scale ranging from 1 to 5, depicted with standard photographs. Tumor areas with 0 to <0.3% vessels were graded as 1; areas with 0.3 to <1% vessels were graded as 2; areas with 1 to <2% vessels were graded as 3; areas with 2 to <5% vessels were graded as 4, and areas with 5% vessels or greater were graded as 5. The density gradings of the three examiners showed excellent agreement (weighted kappa = 0.83) and were averaged for subsequent analyses. Only vessels within the body of the tumors were analyzed. A difference was noted between tumors and normal choroid based on DAPI-stained morphology. Using this difference allowed us to exclude normal choroid from the analysis.

Fig. 1.

Distribution of neovessels and mature vessels. A Summary of the distribution of neovessels and mature vessels. The density of neovessels (CD105) is significantly higher than that of mature blood vessels (α-SMA) in central areas of the tumor. Areas are indicated as follows: apex (a), center (b), base (c), periphery (d). DAPI is the blue nuclear stain. B Neovessels (CD105) are distributed throughout the tumor. C Mature tumor blood vessels (α-SMA) are mostly at the tumor base. D Merged image showing both neovessels (CD105, green) and mature blood vessels (labeled with α-SMA, red).

Statistical Methods

The association of neovessels, mature vessels, and cell proliferation densities with tumor area (i.e., base, center, apex, peripheral edges) was examined with Huynh-Feldt-corrected repeated measure analysis of variance and post-hoc paired t-tests. There was no indication that the data failed the sphericity assumption. The number and percent of tumors with highly aggressive PAS staining were calculated for each tumor area. The average neovessel and mature vessel densities were compared within each tumor area with respect to highly aggressive and less aggressive PAS staining using a two-factor analysis of variance, following assessment of test assumptions using Levene's test and residual plots. The relationship between mature tumor blood vessels and known interval level prognostic factors was assessed with Pearson correlation coefficients, while analysis of variance was used for categorical prognostic factors. Results were considered significant if p ≤ 0.05.

Results

Immunofluorescence analysis demonstrated that both neovessels and mature vessels were evident in all tumors, as determined by CD105, α-SMA, lectin, and collagen type IV staining (data not shown). The density of mature vessels was significantly elevated (p = 0.042) in certain areas of the tumor, while the density of neovessels was equal throughout the tumor (p = 0.306; table 2). The density of mature vessels was significantly higher in the basal areas than in the center (p = 0.018), and borderline significantly higher in the apex (p = 0.083).

Table 2.

Comparison of the mean (±SD) grades of neovessels and mature vessels

Location	Mature vasculature (α-SMA)	Neovessels (CD105)	p valuea
Apex	2.6 ± 1.4	3.4 ± 1.0	0.17
Center	2.5 ± 0.9	3.3 ± 1.0	0.036
Periphery	3.1 ± 0.9	3.4 ± 1.0	0.31
Base	3.4 ± 1.6	3.1 ± 1.0	0.47
p valueb	0.042	0.306

There was a statistically significant interaction between areas and degree of vessel maturation (p = 0.044, two-factor repeated measures analysis of variance test of interaction).

^aPaired t tests comparing neovessel and mature vessel density within each area.

^bOne-way repeated measures comparison of vessel density within each vessel type.

The density of neovessels versus mature vessels was analyzed in each tumor region. Neovessel density was significantly greater (3.3 ± 1.0) than mature vessel density (2.5 ± 0.9) in central areas of the tumor (p = 0.036). The neovessel density was also greater than the mature vessel density in apical (3.4 ± 1.0, 2.6 ± 1.4) and peripheral areas (3.4 ± 1.0, 3.1 ± 0.9), but this was not statistically significant (p = 0.17 and p = 0.31, respectively). The mature vessel density was greater (3.4 ± 1.6) than neovessel density (3.1 ± 1.0) in basal areas, but this was also not a significant finding (p = 0.47) (table 2; fig. 1).

To determine the pattern of vessel maturation, the difference between neovessels and mature vessel densities was obtained for every tumor area. The differences obtained for the apical (−0.8 ± 1.9) and central areas (−0.8 ± 1.3) of the tumor were significantly higher than the difference obtained for the basal area (p = 0.014 and p = 0.012, respectively). In contrast, the difference obtained for the peripheral areas (−0.3 ± 1.1) of the tumor was not significantly different from the difference obtained for the base (p = 0.14). The differences obtained for the apical and central areas of the tumor were similar to each other (−0.8 ± 1.9 and −0.8 ± 1.3, least significant difference post-hoc test, table 3).

Table 3.

Comparison of neovessels and mature vessel differences between tumor areas and tumor base

Location	Difference mature vasculature – neovessels (SD)	LSD post-hoc test for difference
Apex	−0.8 (1.9)	p = 0.014a
Center	−0.8 (1.3)	p = 0.012a
Periphery	−0.3 (1.1)	p = 0.14a
Base	0.3 (1.6)

Standard deviation is shown in parentheses.

^aEach area compared to base with least significant difference (LSD) post-hoc test.

To further examine vessel patterns within the tumor, we used data from table 2 to calculate the ratio of mean neovessels to total vessels (i.e., neovessels + mature vessels) and obtained a percentage value for each area: base 47.7%, periphery 52.3%, center 56.7%, and apex 56.9%. This pattern of vessel maturation is similar to that seen with difference analysis. A model of tumor growth suggested by these percentages is presented in figure 2.

Fig. 2.

Model of vessel maturation in uveal melanoma. The developmental pattern of vessel maturation shows a linear and significant increase of the ratio of neovessels to total vasculature, from basal areas of the tumor to leading edges of the tumor. The graph displays the ratio of neovessels to total (neovessels + mature vessels) vessels by tumor area. Inset: Representation of an eye indicating the model of total vasculature pattern of tumor progression.

To test the hypothesis that tumor vessel maturation is correlated with poor prognostic factors, neovessels and mature vessel densities were compared to aggressive PAS patterns, tumor size, epithelioid cell type, extraocular extension, anterior location, mitotic figures, and cell proliferation. No significant associations were found between tumor vasculature and any of the following variables: aggressive PAS patterns, epithelioid cell type, extraocular extension, or anterior location. Associations were found between total tumor vasculature and tumor size, mitotic figures, and cell proliferation. Interestingly, these associations were greater with mature vessel density than with neovessel density (table 4; fig. 3). There was a borderline association between mature vessel density and tumor size (r = 0.48, p = 0.084), but not with neovessel density (r = 0.11, p = 0.72). In addition, mature vessel density was associated with cell proliferation density levels (r = 0.62, p = 0.042), while neovessel density was not (r = 0.29, p = 0.32). Both mature vessels and cell proliferation densities were associated with mitotic figures (r = 0.76, p = 0.001 and r = 0.72, p = 0.013, respectively), while neovessel density was not (r = 0.13, p = 0.67). Taken together these data suggest that larger tumors have a higher cell proliferation and a higher ratio of mature to neovessel density than smaller tumors.

Table 4.

Pearson's correlation coefficients for mature vessels, neovessels and cell proliferation, factors associated with poor prognosis

Prognostic factors	Mature vasculature (α-SMA, n = 14)	Neovessels (CD105, n = 14)	Cell proliferation (Ki-67, n = 11)
Tumor size	r = 0.48 (p = 0.084)a	r = 0.11 (p = 0.72)	r = 0.25 (p = 0.45)
Cell proliferation	r = 0.62 (p = 0.042)b	r = 0.29 (p = 0.32)	–
Number of mitotic figures	r = 0.76 (p = 0.001)b	r = 0.13 (p = 0.67)	r = 0.72 (p = 0.013)b

^aBorderline significant correlation.

^bStatistically significant p value.

Fig. 3.

Correlation between mature vessel density and factors associated with poor prognosis. Graphs showing correlations between mature tumor blood vessels and tumor burden, given as the ratio of the tumor and globe areas (r = 0.48, p = 0.084) (A), cell proliferation density (Ki-67, r = 0.62, p = 0.042) (B), and number of mitotic figures (r = 0.76, p = 0.001) (C).

Discussion

The survival of growing tumors requires the formation of new blood vessels from preexisting ones [23]. Angiogenesis is a complex and highly dynamic process involving a series of events and a multitude of regulatory factors, many of which can be targets for therapy. Evidence suggests that increased expression of CD105, a homodimeric membrane glycoprotein expressed in ECs and upregulated during angiogenesis in the tumor microenvironment [41, 42], is a major and independent biological marker for poor prognosis in a wide number of human tumors including uveal melanoma [18, 40,43,44,45,46,47,48,49,50,51,52,53,54,55]. However, vessel maturation plays an important role in the later stages of tumor development, and experimental studies are lacking on the presence and implication of vessel maturation in human uveal melanoma.

Neovessel formation is inhibited by the use of antiangiogenic agents. ECs in neovessels require growth factors for survival. In their absence, ECs undergo apoptosis and regress [21]. In contrast, mature tumor blood vessels are stabilized by pericytes and are no longer dependent on angiogenic stimuli. The process of vascular maturation results from an intricate association between angiogenic stimulators and inhibitors [56, 57]. Pericytes are recruited to blood vessels at the time at which angiogenic stimuli halt in the tumor. For instance, blood flow and oxygen supply increase immediately following maturation and overlap with decreased levels of VEGF [56, 57]. Mature tumor blood vessels persist for months in spite of angiogenic factor withdrawal [58]. When pericytes and smooth muscle cells are recruited, vessels stabilize and ECs are less likely to undergo apoptosis, becoming more resistant to antiangiogenic treatments [21, 31, 57, 59, 60]. This suggests that vessel heterogeneity, in particular the number of mature vessels found in tumors, may limit the efficacy of vessel targeting therapy as previously reported [57, 61]. In addition, vasculogenic mimicry, in which vessels are lined by tumor cells rather than ECs, may limit the effectiveness of vessel targeting therapy. In vitro data indicate that these extravascular channels do not respond to classical antiangiogenic agents [62]. Therefore, when given as monotherapy, antiangiogenic drugs may have limited efficacy in the treatment of uveal melanoma.

Mitotic figures [7, 8], cell proliferation [18], and tumor size [9, 10] have all been found to be indicators of poor prognosis and death due to metastasis in melanoma. In this paper, we show that the level of vessel maturation is significantly correlated with the number of mitotic figures and cell proliferation in the tumor. An early trend was found between vessel maturation and tumor size at the time of enucleation. Although high levels of neovessel density have been shown in previous studies to correlate with increased mortality [3,4,5], in our study mature vasculature was found to correlate better than neovessel density with other poor prognostic indicators (i.e., mitotic figures, tumor size, and cell proliferation).

Our data indicate that significant differences exist in the spatial distribution of neovessels versus mature vessels in human uveal melanoma. When comparing the density of neovessels from one area to another within the tumors, we found that the density of neovessels is equivalent in all areas of the tumor, while the density of mature vessels is mainly concentrated in the base and periphery of the tumor. When comparing the difference of neovessels to mature vessels within each tumor area, we found that the central, apical, and peripheral areas of the tumors expressed a higher density of neovessels than mature vessels. In contrast, in the basal areas of the tumors, a higher density of mature tumor vessels was present than of neovessels. However, at least a low level of vessel maturation was present in all regions. These mature tumor blood vessels at the base are most likely to be resistant to antiangiogenic treatments [29, 31, 57], and could increase the chance of extraocular tumor extension [4, 11].

Although the choroidal vasculature was excluded in our analysis of mature and neovessels, we believe that the interface between the choroid and the tumor is important in tumor development and tumor vessel maturation. We hypothesize that the pattern of vessel distribution that this study uncovered may be explained by the close proximity of the tumor base to the normal mature vasculature of the choroid, allowing co-optation of already mature choroidal vasculature by the tumor. We believe that the choroidal component of the vasculature initially feeds the tumor at the margins, and as the tumor develops, this bridge to a normal vascular network is gradually lost at the apex. Another potential explanation is that the basal region of the tumor is believed to be the oldest, allowing the most time for increased maturation of the vasculature at the base of the tumor. Thus, these tumors may have dual vascular components: (1) the original component branching from the choroidal vasculature and (2) the neovascular component at the leading edges of the tumor resulting from angiogenesis.

We initially hypothesized that neovessels would localize to the leading edges of the tumor. The data collected in this study support our hypothesis and allow us to propose a model for vessel development in uveal melanoma. Since the ratio of neovessel to total vessel density by tumor location in the apical and central areas was similar (i.e., apex is 56.7%, and center is 56.9%), both areas were averaged for representational purposes. The developmental maturational pattern of tumor progression shows a linear and significant increase in the ratio of neovessels to total vasculature, from basal areas of the tumor to leading edges of the tumor (fig. 2). This model provides a potential understanding for spatial tumor vessel maturation associated with progressive tumor development.

Pericytes associated with mature tumor blood vessels have gained new attention as functional and critical contributors to tumor angiogenesis and potential new targets for therapy [29]. Because uveal melanoma detection usually occurs during later stages of tumor development [63,64,65], significant vessel maturation is present. Thus an ideal therapy would include vascular targeting agents which target neovessels as well as agents which target mature vessels such as pericyte-targeting agents [29]. Such a strategy would provide a comprehensive, integrative, therapeutic approach for the treatment of human uveal melanoma.