Monosomy 3 Predicts Death but Not Time until Death in Choroidal Melanoma

Abstract

purpose. To study whether monosomy 3 can predict time until death caused by metastatic melanoma, whether life expectancy can be predicted in patients after surgical excision of a melanoma displaying monosomy 3, and to confirm the prognostic value of monosomy 3 and its correlation with tumor histology.

methods. Archival specimens from 71 patients who died of metastatic melanoma and 40 patients who were living or had died of other causes were identified. The number of copies of chromosome 3 was assessed by chromosome in situ hybridization, and monosomy 3 was compared with clinicopathologic features.

results. Monosomy 3 was detected in 47 of 71 metastasizing melanomas (66.1%) and was significantly associated with metastasis-related death (P < 0.0001). All 40 nonmetastasizing tumors were balanced for chromosome 3 (two copies). In 70% of cases, epithelioid cells and vascular loops in combination predicted the presence of monosomy 3 (P < 0.0001). Among the 71 patients who had died of metastasizing melanoma, there was no difference in time until death between monosomic and balanced tumors. However, a survival curve corrected for age of the patients at the time of surgery suggested that very-long-term survival with monosomy 3 is probably rare.

conclusions. Monosomy 3 is an important predictor of death in melanoma and is in some cases predicted by histology. However, death of metastatic disease occurs in a significant number of patients without monosomy 3. There is no significant difference in time until death between metastatic melanomas, with and without monosomy 3. However, survival of patients with tumors displaying monosomy 3 is generally short.

Uveal melanoma remains the most common primary intraocular malignancy in adults, with an overall mortality of approximately 50% because of the development of liver metastases. ^{1 2} These metastases are not usually evident when the primary tumor is treated, and much current research has been conducted to identify reliable prognostic markers in primary uveal melanoma for early detection of high-risk patients in whom prophylactic treatment would be justified. Specifically, monosomy 3 has been recognized to have a stronger association with a poor prognosis than do clinical and histologic parameters and is in some series associated with a reduction in the 5-year survival time from almost 100% to only 30%. ³  

It is generally accepted that the peak mortality from metastatic disease occurs within 3 years of diagnosis. ² However, the clinical course of patients with uveal melanoma is unpredictable, and a significant number of deaths still occur after 5, 10, and 15 years. The occurrence of very late metastases up to 36 years after initial surgery has also been described. ⁴ To date, the role of monosomy 3 in predicting time until death has not been assessed. 

The purposes of this study were to confirm the prognostic value of monosomy 3 in predicting death caused by metastasis and to assess whether the presence of monosomy 3 can be predicted from tumor histology. We then assessed whether monosomy 3 is related to time until death caused by metastasis and whether life expectancy can be predicted in patients after surgical excision of a melanoma displaying monosomy 3. 

Materials and Methods

Case Selection

Archival specimens from patients who had choroidal melanoma were obtained from the Glasgow Western Infirmary Pathology files. A total of 111 samples were obtained, including 60 enucleations, 50 local resections (LRs), and 1 exenteration. The patients were divided into two groups: group 1 (patients who had died of metastatic disease) and group 2 (patients who were living or had died of other causes after a minimum follow-up period of 1 year; mean, 16.85 years; range, 1–30). There were 71 cases in group 1, metastasizing melanoma (MM), and 40 cases in group 2, nonmetastasizing melanoma (NMM). Follow-up time was recorded from date of treatment to death or to last follow-up. The surgery was performed between 1974 and 1992 in patients with MM and between 1973 and 1991 in patients with NMM. Patients with MM were identified either from the cancer registry or case notes as having definite liver metastases that were identified by imaging, biopsy, or postmortem examination. Patients with NMM were alive and well or had a cause of death other than metastatic melanoma and no evidence of metastatic disease at last follow-up. All tissues had been fixed in glutaraldehyde or formalin and embedded in paraffin wax. In 15 cases from both groups, the nature of the fixative was not known. Sixteen of the 71 specimens of MM and four of the 40 cases of NMM had been fixed in formalin (FF). The remaining 53 MM and 35 NMM specimens had been glutaraldehyde fixed (GF). The FF and GF specimens were of similar ages, and tissue age and fixation has been shown not to affect results in chromosome in situ hybridization (CISH). ⁵ This project received the full approval of the West Ethics Committee, North Glasgow Trust, and adhered to the tenets of The Declaration of Helsinki. 

Clinical Features

Clinical details were obtained from case notes and the cancer registry. The details sought included age, sex, treatment of tumor, survival status, and cause of death. The time until death was calculated from the date of LR or enucleation. Survival status was determined up to August 2003. After these clinical details were obtained, all samples were rendered anonymous. 

Histologic Features

Sections (4 μm) were cut and stained with hematoxylin and eosin (H&E) and periodic acid-Schiff (PAS) without counterstain for light microscopy. The following histopathologic parameters were assessed according to the Collaborative Ocular Melanoma Study ⁶ : cell type, extent of necrosis, maximum tumor dimension, intrascleral invasion, extraocular extension, degree of pigmentation, mitotic activity, and lymphocytic infiltration. Closed vascular loops were assessed in PAS-stained sections with the aid of a green filter, as described by Folberg et al. ⁷ and Foss et al. ⁸ The extent of retinal detachment (RD) was classified into one of three categories: 1, no RD; 2, focal RD: detachment adjacent to tumor (detachment of neurosensory retina over and around the melanoma) or on the same side of tumor but not adjacent; and 3, total RD: detachment of neurosensory retina with three to four quadrants of the retina involved. The exenteration specimen and the cases of local excision could not be assessed for RD. 

Chromosome In Situ Hybridization

The number of copies of chromosome 3 was assessed by CISH with chromosome-specific centromeric probes, as previously described. ⁵ Chromosome 18 was used as a control chromosome, as it rarely shows abnormalities in uveal melanoma. In 20 enucleation specimens (7 FF and 12 GF), normal retina was used as an internal control. In addition, 19 cases of normal human skin (all GF) were included as an external control. Briefly, 4-μm sections were pretreated by microwaving followed by digestion in pepsin (0.4% pepsin in 0.2 M hydrochloric acid) for 30 minutes at 37°C. The tissue sections and probe were then simultaneously denatured for 5 minutes at 80°C, to obtain single-stranded DNA, and hybridized overnight at 37°C. Sites of hybridization were detected by using anti-digoxigenin alkaline phosphatase (AP) Fab fragments (Roche, Indianapolis, IN). Sites of hybridization and assessment of the number of chromosomes were performed as previously described. ⁵ In each sample, 200 cells were counted. Poor quality hybridization signals and overlapping nuclei were not analyzed. Identification of monosomy 3 was performed by quantifying the chromosome index (CI) and signal distribution (SD) within the tumor and comparing with normal tissue (retina). The CI gives an average chromosome copy number and is calculated by dividing the number of hybridization sites by the number of nuclei counted. Chromosome loss was defined as a CI less than 3 standard deviations from the mean for retina. The standard deviation was defined as the percentage of nuclei with one hybridization site. An SD of more than 60% was the cutoff point used to define chromosome loss. Tumors had to show chromosome loss by both CI and SD, to be regarded as monosomic. 

Statistical Analysis

The relationship between MM and age, mitotic activity, tumor size, and pigmentation was assessed by the Wilcoxon rank sum test. For the other prognostic variables (monosomy 3, sex, cell type, intrascleral and extraocular spread, lymphocytic infiltration, presence of vascular loops, necrosis, and RD), associations were assessed by the Fisher exact test. The probability was corrected for the multiplicity of tests by Holm’s method. ^{9 10} Holm’s method ensures that there is no false conclusion, compensating for the multiplicity of statistical tests. Multivariate analysis of several risk factors was based on a stepwise discriminant analysis, as implemented in a statistical-analysis software package (SPSS; SPSS, Inc., Chicago, IL). Survival time of a patient after surgical excision of a melanoma with monosomy 3 was compared with that of those with tumors balanced for chromosome 3 (two copies), by using the two-sample t-test. The life expectancy of patients with tumors displaying monosomy 3 was predicted by using a survival curve corrected for the patient’s age. 

Results

Association of Monosomy 3 with MM

Monosomy 3 was detected in 47 of the 71 cases of MM (66.1%) and was significantly associated with metastasis-related death (P = 0.0001). The remaining 24 MM and all 40 NMM had two copies of chromosome 3 and therefore were defined as balanced. 

Prediction of Monosomy 3 from Clinical and Histologic Features

The clinical and histologic features are summarized in Table 1 . Multivariate discriminant analysis of the histologic features of both MM and NMM indicated that epithelioid cells and vascular loops were statistically significant predictors of monosomy 3, when considered together. Both satisfied the confidence criterion P < 0.0001. The discriminant function derived gave the correct prediction of chromosome number in 70% of cases. 

Prediction of Survival Time in Tumors with Monosomy 3

Among the 71 patients in group 1 (MM), 47 displayed monosomy 3, with a mean survival time of 4.31 ± 2.83 years (SD). The remaining 24 with two copies of chromosome 3 had a mean survival time of 4.21 ± 3.34 years. There was no significant difference in time until death of those with tumors with monosomy 3 compared with those with two copies of chromosome 3 in this group (P = 0.90; Fig. 1 ). 

Prediction of Life Expectancy of Patients after Surgical Excision of a Melanoma Displaying Monosomy 3

All 47 patients with melanomas displaying monosomy 3 had died. The mean survival time was 4.3 years (median, 3.5). The survival curve, corrected for the patient’s age at the time of surgery and based on the assumption of a linear relationship, suggests that very-long-term survival with monosomy 3 is rare (Fig. 2) . 

Discussion

Monosomy 3 has been shown by many investigators to be associated with death caused by metastasis after treatment of uveal melanoma. ^{3 11 12 13 14 15 16 17 18} Using the technique of CISH, ⁵ which is applicable to archival tissue, we confirmed the findings of these previous studies. In this study, we identified monosomy 3 in 66.1% of cases of melanoma with death caused by metastasis. Previous studies have identified monosomy 3 in 50% to 73% of choroidal melanomas. ^{3 12 13 14 17 18} It is not possible to compare these data directly, as, unlike some of these studies, our data set was selected on the basis of death or survival in patients with uveal melanoma and is therefore not considered to be as representative of all cases of choroidal melanoma. 

Although we have confirmed that monosomy 3 is a significant predictor of metastasis-related death, we have identified a second smaller but significant group of patients who died of metastatic melanoma even though their tumors contained two copies of chromosome 3. There are several possible explanations for this. First, this is a retrospective study that includes samples from the 1970s. We may therefore have included samples from patients who would now be treated without surgery. Without tissue samples, patients with similar tumors could not be included in other studies. These differences may also be related to the CISH technique used in this investigation, since the probes are centromere specific, and regional losses of chromosome 3 would not be detected. However, in previous studies in which other techniques were used, including microsatellite analysis and comparative genomic hybridization, investigators found that in most cases the entire copy of chromosome 3 is lost and that regional loss occurs in only a minority. ^{3 14 15 18 19} Furthermore, the prognostic significance of these regional losses is unclear. Scholes et al. ¹³ identified loss of heterozygosity of chromosome 3 in 60 cases; of these, 6 cases showed only regional losses. Five of these six patients were alive at the end of the 4-year study period, suggesting that these partial deletions may not carry the prognostic significance of complete loss of heterozygosity. It is also possible that we have failed to identify a small subclone of cells within the tumor displaying monosomy 3. However, the presence of subclones of cells is more likely to be identified by using CISH compared with standard cytogenetics, as a whole histologic section of the tumor is studied and not just selected cells that have been successfully cultured and identified in a metaphase spread. Finally, and probably most likely, there may be other alterations that are important in the development of metastases in choroidal melanoma. These may include other as yet unidentified cytogenetic abnormalities or may be related to alterations in tumor biology that allow invasion and metastasis. The metastatic cascade involves a series of events including detachment of tumor cells from each other followed by both attachment to and degradation of the extracellular matrix, thus allowing migration of tumor cells. Alterations in cell adhesion molecules have been shown to correlate with an increased risk of metastasis in choroidal melanoma. For example, loss of intercellular cell adhesion molecule-1 expression has been shown to be associated with an increased risk of metastasis within the first 5 years of diagnosis. ²⁰ Conversely, Woodward et al. ²¹ showed that invasive melanoma cell lines express α1- and α4 -integrins that are not expressed on noninvasive cell lines. Furthermore, these invasive cell lines displayed better adhesion to extracellular matrix substrates and endothelial cells than their noninvasive counterparts. MMs have also been shown to display increased expression of matrix metalloproteinase-2 that may play a role in degrading the extracellular matrix. ²² The molecular mechanisms that lead to the metastatic phenotype in choroidal melanoma are not fully understood but may be related to isolated point mutations in relevant genes or their promoters, so that they would not be detected in chromosomal studies. 

In this retrospective study, there were 71 patients who died of metastases, with time from surgery to death ranging from 4 months to 14 years; two thirds of these had tumors displaying monosomy 3 and a mean survival time of 4.31 years; the remaining third were apparently balanced for chromosome 3 and had a mean survival time of 4.21 years. There was no significant difference in time until death between the two groups. This study also showed that survival beyond 10 years is possible but rarely occurs in patients with a tumor displaying monosomy 3. Because some of the patients were identified from a database in which they were only included after death, it is possible that very-long-term survivors are underrepresented in the data, with consequent underestimation of survival time. However, using a survival curve to predict the life expectancy of these patients suggests that very-long-term survival with monosomy 3 is probably rare, and as such this bias is small. 

In this study, the presence of monosomy 3 was partly predicted from tumor histology. Specifically, the presence of epithelioid cells (P < 0.0001) and closed vascular loops (P < 0.0001) were significantly associated with monosomy 3, when each characteristic was considered individually. When these two histologic characteristics were included in a multivariate predictive model, the predictions were correct in 70% of cases. Scholes et al. ¹⁴ also showed a significant association between monosomy 3, closed vascular loops, and epithelioid cells; however, by using a forward stepwise logistic regression model, they showed that monosomy 3 could only be reliably predicted in patients with large epithelioid tumors. ¹³ In our study, large tumor size (>15 mm) was not a significant predictor of monosomy 3. The distribution of tumor sizes was similar in this study to that in Scholes et al. ¹³ and it is therefore unlikely to represent sample selection. Other researchers have shown no association of monosomy 3 with histopathologic features. ^{3 15 18}  

In conclusion, our study confirms that monosomy 3 in choroidal melanoma is a significant predictor of metastasis-related death. However, death caused by metastatic melanoma occurs in a significant number of cases that do not display monosomy 3, suggesting that other molecular events are important in the metastatic cascade in these patients. Furthermore, this emphasizes a cautious approach to the use of cytogenetic studies for patient counseling. In this study, the presence of monosomy 3 was correctly predicted in up to 70% of cases by the presence of an epithelioid cell component and vascular loops, but it was not related to tumor size. In patients with metastases, there was no difference in time until death in those with tumors displaying monosomy 3 compared with those tumors that had with two copies of chromosome 3. However, as with all MMs, life expectancy in patients with tumors displaying monosomy 3 is generally short. 

 Presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, April, 2004.

 Submitted for publication May 18, 2005; revised June 25, 2005; accepted August 29, 2005.

 Disclosure: M.T. Sandinha, None; M.A. Farquharson, None; I.C. McKay, None; F. Roberts, 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.

 

Clinical and Histological Parameters of Metastasizing and Nonmetastasizing Choroidal Melanoma

	MM (n = 71)	NMM (n = 40)
Mean age (y)	57.8	52.8
Gender
Male	36	19
Female	35	21
Treatment
Enucleation	37	23
LR	33	17
Exenteration	1	0
Chromosome 3
1 Copy	47	0
2 Copies	24	40
Chromosome 18
1 Copy	0	0
2 Copies	71	40
Cell type
Epithelioid present	57	7
Spindle	14	33
Balloon cells
Yes	7	13
No	64	27
Necrosis
Yes	23	3
No	48	37
Pigment
None to low	36	27
Moderate	25	12
Heavy	8	1
Lymphocytes
None to low	59	38
Moderate	8	2
Heavy	4	0
Invasion
No	32	23
Intrascleral	21	14
Extraocular extension	18	2
Mitoses
<5	53	38
5–10	15	1
>10	3	1
Loops
Yes	39	8
No	32	32
Size (mm)
<1	8	10
11–15	16	15
>15	47	15
Retinal detachment
No	0	1
Focal	11	6
Total	26	9

 Data are the number of cases, unless marked otherwise.

Figure 1.

 

Dot plot showing time from surgery to death of patients with MMs containing one (monosomy 3) or two (balanced) copies of chromosome 3.

Figure 2.

 

Estimation of life expectancy, corrected for age, in patients with MM displaying monosomy 3, based on the assumption of a linear relationship.

The authors thank Jim Ralston for lending technical expertise.