The Spatial Distribution of Monosomy 3 and Network Vasculogenic Mimicry Patterns in Uveal Melanoma

Abstract

purpose. Monosomy of chromosome 3 and network vasculogenic mimicry patterns are associated with death in patients with uveal melanoma (UM). Networks are typically found in confined areas within the tumor, whereas the intratumor distribution of chromosome 3 aberrations is unknown. This study was conducted to assess the spatial correlation among chromosome 3 aberrations and networks in UM.

methods. Vasculogenic mimicry patterns, proliferative activity, and cell type were characterized in 15 enucleated eyes with primary UM. Cells were isolated by laser capture microdissection (LCM) from two tumor regions and one normal retina area from each tissue block. In the eight tumors containing networks, the cells were microdissected from one area with networks and a different area without networks. In seven tumors without networks, cells were microdissected from two distinct tumor areas. The presence of chromosome 3 aberrations was assessed by microsatellite analysis (MSA) in each LCM sample.

results. Useful MSA data was obtained from 43 of the 45 samples. Monosomy 3 was detected in 16 samples of eight tumors. There was no intratumor heterogeneity for monosomy 3, regardless of the existence of heterogeneity in networks, cell type, or proliferative activity across the two samples from the same tumor. Networks were associated with the presence of monosomy 3 throughout the entire tumor (P = 0.02).

conclusions. Of the histologic prognostic factors of metastasis in UM studied, only the presence of a network vasculogenic mimicry pattern but not its location is associated with monosomy 3. This suggests that monosomy 3 may contribute to but is not sufficient for the development of the network pattern.

Several histologic, genetic, and demographic prognostic factors have been identified in uveal melanoma (UM). Among these factors, identification of vasculogenic mimicry patterns in tumor sections, loss of one copy of chromosome 3 (monosomy 3), and specific gene expression patterns have been most significantly associated with death from metastatic disease. ^{1 2 3 4 5 6 7 8 9 10 11} Among other histologic prognostic factors in UM are cellular composition (spindle or epithelioid cells) and tumor cell proliferative activity. ^{7 12 13 14}  

Although nine vasculogenic mimicry patterns have been identified in UM, death from metastatic disease is particularly associated with the presence of the network of loops pattern. This distinct form probably evolves during tumor progression, since several different patterns are usually present across a single tumor section and since such patterns are not homogenously distributed across different tumor areas. ^{1 3 15 16 17} Heterogeneity across different tumor areas has also been observed in cell type composition and proliferative activity. ^{7 12 13 14}  

The most common chromosomal aberration in UM is monosomy 3. It is strongly associated with death from metastasis ^{5 18 19 20 21 22} as well as with gene expression patterns. ¹¹ Consequently, the existence of tumor suppressor genes on chromosome 3 has been suggested, ^{23 24} although no specific gene has been identified. 

In view of current knowledge of their prognostic importance, it seems plausible that prognostic factors such as vasculogenic mimicry patterns, chromosomal aberrations, gene expression pattern, proliferative activity, and cell type are causally related. In accordance with that, correlations between the presence of the loop vasculogenic mimicry pattern and monosomy 3 and between vasculogenic mimicry patterns and gene expression patterns have been described. ^{9 25} However, the distribution of chromosomal aberrations within tumors and its spatial correlation with additional histologic prognostic factors is unknown. 

Defining such spatial correlation among chromosomal aberrations and other histologic prognostic markers may clarify whether these aberrations are an early event in UM development. In such cases, chromosomal aberrations may be required but not sufficient for the development of characteristics such as vasculogenic mimicry patterns associated with unfavorable prognosis. In contrast, lack of any correlation suggests that these prognostic factors are not directly related. To that end, we have studied the spatial correlation among chromosome 3 aberrations and the presence of network vasculogenic mimicry patterns in primary UM. 

Methods

Patients and Tissues

Fifteen formalin-fixed, paraffin-embedded, blocks of primary UM tumors were included in the study (Table 1) . Patients were treated and eyes were enucleated in the Department of Ophthalmology of the Hadassah-Hebrew University Medical Center, Jerusalem, Israel. The study was approved by the Institutional Helsinki Committee and was conducted in accordance with the Declaration of Helsinki. 

Conventional Histology, Periodic Acid Schiff Staining, and Immunohistochemistry

Hematoxylin and eosin (H&E) and PAS stainings were performed on sequential 5-μm paraffin-embedded sections, as previously described. ²⁶ Cell type was determined in H&E-stained sections by an ophthalmic pathologist (JP). Vasculogenic mimicry patterns were identified in PAS-stained sections. ²⁶  

Proliferative activity of tumor cells was assessed in tumor sections which were immunostained for ki-67 using a monoclonal anti human ki-67 antibody (M 7240; DakoCytomation, Glostrup, Denmark), as previously described. ¹⁶ Proliferative index was determined by calculating the mean positive cell count per high-power field (×400) in 10 such fields. 

Laser Capture Microdissection and DNA Extraction

Laser capture microdissection (LCM; Arcturus, Mountain View, CA) was performed on paraffin-embedded sections to isolate at least 1000 cells from two distinct areas in each tumor. Each such area measured approximately 1 to 2 mm in diameter. In the eight tumors that contained network vasculogenic mimicry patterns, the cells were microdissected from one area containing the network pattern and another area that contained either no pattern (silent), normal patterns, or straight patterns. Both areas were separated as far as possible, but did not contain tumor edge or normal tissue. Cells were dissected from the entire area including the vasculogenic mimicry pattern present in the area (Fig. 1) . In the additional seven tumors that did not contained network pattern, cells were microdissected from two distinct tumor areas. These tumor regions were also separated as far as possible from one another. Cell type and proliferative activity were assessed in both areas in each tumor from which cells were microdissected. Cells were also isolated from areas of normal retina tissue in each section. 

Before LCM formalin-fixed, paraffin-embedded sections were deparaffinized and dehydrated in the following order: Sections were placed in xylene for 5 minutes, 100% ethanol for 30 seconds, 95% ethanol for 30 seconds, 70% ethanol for 30 seconds, double-distilled water for 30 seconds, 95% ethanol for 60 seconds, 100% ethanol for 60 seconds, and xylene for 5 minutes and were air dried under a hood for 5 minutes. DNA was extracted by incubation of captured cells in lysis buffer (0.04% proteinase K, 10 mM Tris-HCl [pH 8.0], 1 mM EDTA, and 1% Tween 20; final pH of the buffer was 8.0) at 65°C overnight followed by proteinase K inactivated by incubation in 95°C for 8 minutes. 

Microsatellite Analyses

Chromosome 3 aberrations were studied by microsatellite analysis with fluorescence-labeled primers as previously described ²⁷ (model 3100 sequence detection system and GeneScan and Genotyper software; ABI, Foster City, CA). Briefly, polymorphic microsatellite loci on chromosome 3 (D3S3050 on 3p26, 3.2 Mb; D3S2406 on 3p13, 73 Mb; D3S3045 on 3p12-q13, 105.5 Mb; D3S1744 on 3q24, 145 Mb; and D3S1311 on 3q29, 195 Mb) were at first assessed in DNA isolated from section areas not affected by the tumor (normal retina). Because of the limited amount of DNA extracted from LCM samples one or two informative markers were analyzed in the corresponding tumor samples. 

Results

LCM and DNA extraction were successfully applied to isolate a total of 45 samples from 15 eyes (Fig. 1) . From each eye there were two tumor samples and one normal retina sample. Microsatellite analysis (MSA) from the LCM samples enabled detection of chromosome 3 status in 43 of the 45 samples (Fig. 2) . Two samples, one from a tumor (Table 2 , sample 3b,) and one from normal retina (sample 15) did not yield useful MSA information. Although sample 15 probably contained monosomy 3 in both tumor areas, such monosomy could not be confirmed, as MSA of the normal retina from this sample failed, therefore, this sample was not included in the analysis. Table 2presents histologic findings, proliferative activity, vasculogenic mimicry pattern classification, and chromosome 3 status in the 14 informative tumors. 

Eight of the 14 informative tumors showed monosomy 3. There was no intratumor heterogeneity for monosomy 3 between the two areas that were evaluated, regardless of whether there was heterogeneity between these areas in the presence of network patterns (in seven tumors), cell type (in five tumors), or proliferative activity (in five tumors). 

Each of the seven informative tumors that showed network vasculogenic mimicry patterns also showed monosomy 3. There were only two tumors in which monosomy 3 was detected, despite the absence of a network pattern from the tumor. Thus, at the whole tumor level, network pattern was associated with monosomy 3 (P = 0.02, Fisher’s exact test). By contrast, there was no association at the whole tumor level between tumor location, tumor height, tumor size, presence of ki67-positive cells, or cell compositions and monosomy 3 (Table 3) . It is noteworthy that the study was not design to exclude such associations, and thus, it is still possible that such association between these prognostic factors and monosomy 3 exist. 

Discussion

This study demonstrates an association between the presence of monosomy of chromosome 3 and the network vasculogenic mimicry pattern on the whole tumor level. Yet, although the network pattern is usually confined to an area within the tumor, chromosome 3 aberrations were also present in tumor regions without networks. Similarly, there was no spatial correlation between proliferative activity or intratumoral heterogeneity of cells and the presence of chromosome 3 aberrations. 

These data suggest that chromosome 3 aberrations probably precede the development of intratumor heterogeneity in vasculogenic mimicry patterns, cell composition, and proliferative activity in UM. In accordance with that, Prescher et al. ²⁸ suggested that chromosome 3 aberrations are an early event in UM based on cytogenetic analysis of two subclones from a single UM tumor. Furthermore, Maniotis et al. ^{29 30} have demonstrated that highly invasive uveal melanoma cells are essential for the formation of patterns in vitro. These data along with the high correlation between the presence of chromosome 3 aberrations and network patterns on the whole tumor level that we and others have identified ²⁵ suggest that chromosome 3 aberrations may be important but not sufficient for the development of the network pattern. Further studies are necessary to clarify whether monosomy 3 is a prerequisite for network pattern formation in vivo. 

Recently, the use of fine needle aspiration (FNA) has been suggested to enable prognosis assessment based on the identification of chromosome 3 aberrations in UM samples. ³¹ Frequent intratumor heterogeneity in chromosome 3 status would significantly impair metastatic risk assessment based on FNA in UM patients. Our study was not designed to test for such general heterogeneity, as we dissected only two areas from each tumor. However, it is worth mentioning that although our results do not reflect such general intratumor heterogeneity in chromosome 3 aberrations, a recent study suggested that such heterogeneity may exist in UM and that it may correlate with areas of discrete epithelioid cell composition. ³² Further studies designed specifically for assessment of chromosome 3 heterogeneity across uveal melanoma, as well as studies correlating analysis of chromosome 3 status in FNA and tumor sections are needed to evaluate this potential diagnostic tool. 

 Supported by a Clinical Research Career Development Award (IC) and a postdoctoral fellowship award (TM) from the Israel Cancer Research Fund (ICRF), by a research grant from the Israel Cancer Organization, and by Deutsche Forschungsgemeinschaft (DFG) Grant LO530/3-5-KFO109.

 Submitted for publication October 31, 2006; revised December 13, 2006, and January 3, 14, and 28, 2007; accepted March 14, 2007.

 Disclosure: T. Meir, None; M. Zeschnigk, None; L. Maßhöfer, None; J. Pe’er, None; I. Chowers, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Patients’ Demographics, Tumor Size, and Tumor Location

Tumor	Age	Sex	Larger Basal Diameter (mm)	Tumor Height (mm)	Location
1	35	M	18	13	Ciliary body and choroid
2	50	F	17	8	Ciliary body
3	67	M	19	2	Ciliary body and choroid
4	57	F	11	6	Ciliary body
5	82	M	18	14	Ciliary body
6	84	M	13	9	Ciliary body and choroid
7	56	M	20.5	12.5	Choroid
8	75	F	16	16	Choroid
9	76	M	12.5	15	Choroid
10	61	F	14	11	Ciliary body and choroid
11	79	M	21	11	Ciliary body and choroid
12	70	M	15	9.5	Choroid
13	69	F	20	18.5	Ciliary body and choroid
14	63	M	9	2	Ciliary body and choroid
15	51	F	11	9	Choroid

Figure 1.

 

LCM from UM sections. Tumors were characterized for vasculogenic mimicry patterns in PAS sections, and cells were microdissected from areas characterized by the presence of a network pattern (A, area delineated by an ellipse, arrow), or absence of such a pattern (B, area delineated by an ellipse). Both areas in (A) and (B) are from the same tumor. Spaces are apparent in the UM section from the same tumor area in (A) after the microdissection of cells (C). Magnification, ×40. Bar, 1 mm.

Figure 2.

 

MSA on DNA extracted by LCM. Loss of heterozygosity in both samples extracted from one tumor is evident by a single peak for the marker D3S3045 (A). Retention of heterozygosity for the marker D3S1744 is seen in both tumor samples from another tumor (B). (+) and (−): presence or absence of network pattern, respectively.

Table 2.

 

Histological Characteristics and Chromosome 3 Status in LCM Samples

Tumor	LCM-Sample	Cell Type	Proliferative Activity	Network Patterns	Monosomy 3
1	a	Epithelioid	15	+	+
	b	Epithelioid	−	−	+
2	a	Spindle	−	+	+
	b	Spindle	−	−	+
3	a	Spindle	1	−	+
	b	Mixed	−	−	Failed
4	a	Spindle	15	+	+
	b	Spindle	16	−	+
5	a	Spindle	7	+	+
	b	Spindle	16	−	+
6	a	Mixed	110	−	+
	b	Spindle	142	−	+
7	a	Mixed	11	+	+
	b	Mixed	10	−	+
8	a	Epithelioid	−	−	−
	b	Spindle	−	−	−
9	a	Epithelioid	−	+	+
	b	Spindle	−	−	+
10	a	Epithelioid	−	+	Partial monosomy
	b	Spindle	−	−	Partial monosomy
11	a	Epithelioid	112	−	−
	b	Spindle	−	−	−
12	a	Epithelioid	13	−	−
	b	Epithelioid	−	−	−
13	a	Epithelioid	100	−	−
	b	Epithelioid	−	−	−
14	a	Spindle	2	−	−
	b	Spindle	−	−	−

 Proliferative activity, mean ki67-positive cell count for high-power field with (−) indicating no ki67 positive cells; Network Pattern, (+) and (−), presence or absence of network pattern in the LCM sample, respectively; partial monosomy 3, D3S1744 (3q24) shows lack of heterozygosity, D3S3050 (3p26) shows retention of heterozygosity, suggesting a deletion of parts of 3p only.

Table 3.

 

Summary of Statistical Analysis for Assessment of Correlation among Histological Prognostic Factors and Monosomy 3 throughout the Tumor

Prognostic Factor	Monosomy 3			P
		Present	Absent
Tumor height (mm)	10.3 ± 4.4	11 ± 5.8	0.9
Larger basal diameter (mm)	15.7 ± 3.5	15.3 ± 4.8	0.8
Proliferative activity (number of tumors with/without ki-67+ cells)	6/3	4/1	1
Cell composition (number of tumors composed of spindle/epithelioid cells)	3/6	1/4	1
Tumor location (number of tumors located at ciliary body/choroid)	6/2	3/3	0.58
Number of tumors with/without networks	7/2	0/5	0.02

 For tumor size the mean height and dimension were compared between tumors with and without monosomy 3, respectively, by t-test. The potential association of histologic prognostic factors with monosomy 3 was evaluated by correlating the relative number of tumors with ki67⁺ cells, epithelioid cells, and networks with the identification of monosomy 3, by the Fisher exact test.