Proportion of microsatellite instability (MSI)–positive cases with MSI at specific microsatellite loci.
Microsatellite instability in choroidal melanomas. Arrowheads indicate the microsatellite instability bands. N indicates normal; T, tumor.
Mismatch repair protein expression. A, Absence of hMLH1 expression in the tumor cells (negative control). B, Uniform hMLH1 nuclear expression in the cells of the inner and outer nuclear layers of the retina (positive control). C and D, hMSH2 protein expression in the nuclei of the tumor cells.
Kaplan-Meier plot of survival: low-level microsatellite instability vs microsatellite-stable subsets.
Hussein MR, Haemel AK, Albert DM, Wood GS. Microsatellite Instability and Alterations of Mismatch Repair Protein Expression in Choroidal Melanomas. Arch Ophthalmol. 2005;123(12):1705-1711. doi:10.1001/archopht.123.12.1705
WIGGSJANEY L.MD, PhD
Copyright 2005 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2005
To examine choroidal melanomas for genomic instability, manifested by microsatellite instability (MSI) and mismatch repair (MMR) protein alterations, and to determine the association of these alterations with selected clinicopathological features of the tumors.
Polymerase chain reaction–based microsatellite assays were applied to analyze 57 cases of choroidal melanomas using 11 microsatellite markers at 5 chromosomal regions: 1p, 2p, 4q, 9p, and 17p. Immunoperoxidase staining methods and mouse monoclonal antibodies were used to investigate the expression patterns of MMR proteins.
Microsatellite instability was found at the 1p, 9p, and 17p regions in these lesions with an overall prevalence of 35% (20/57). The frequency of MSI ranged from 9% (1/11) to 27% (3/11), ie, low-level MSI (MSI-L). The instability was most commonly found at the 1p region (D1S2734, 55%; D1S2832, 40%; and D1S233, 20%). Two MSI banding patterns, band shifts and the appearance of additional bands, were found in 10% and 90% of the unstable lesions, respectively. The average percentages of hMLH1 and hMSH2 positively stained cells were insignificantly reduced in the unstable lesions (81.7 ± 9.3 and76.7 ± 16.7) as compared with the stable lesions (84.1 ± 15.5 and 78.6 ± 19.6; P = .62 and 0.74 for hMLH1 and hMSH2, respectively). There was no significant difference in survival between the 2 groups; however, relative to the stable subset, the unstable tumors showed a trend (P<.10) toward occurring at a younger age and having tumor cells in vascular lakes.
The presence of MSI-L in some choroidal melanomas defines a novel genetic subset of these tumors and suggests that MSI (genomic instability) may play a role in their molecular pathogenesis. Elucidation of the underlying mechanisms for MSI will require further investigation.
Detection of the MSI-L pattern might prove to be useful as an adjunct to the conventional diagnosis of choroidal melanomas. Larger series are needed to determine whether any of the correlative trends noted in this study will achieve statistical significance. To the best of our knowledge, this study is the first to define both the MSI and MMR protein expression features of choroidal melanomas.
The calamities of cancer and blindness are combined in uveal melanoma, the most common primary intraocular malignancy in white adults. These tumors originate from the melanocytes of the uveal tract and have high metastatic tendencies.1- 3
Microsatellites are repetitive DNA sequences that show the same fixed length in all of a given individual’s different tissues. The variation in microsatellite length between tumors and their matching normal tissues is referred to as microsatellite instability (MSI).4 This instability may be associated with defective DNA repair mechanisms that have a negative impact on the fidelity of replication in a more generalized way, with an increased rate of mutations in oncogenes and tumor suppressor genes.4 According to the level of instability, tumors with MSI are categorized into 2 groups. Those with instability at greater than 30% of tested markers, ie, a high instability (MSI-H) pattern, are associated with mismatch repair (MMR) gene defects.4 The other group, with still unknown underlying mechanisms, has instability at less than 30% of tested markers, ie, a low instability (MSI-L) pattern. The MMR system repairs errors during DNA replication, and its components include the human homologues 1, 2, and 6 (hMLH1, hMSH2, and hMSH6). These genes are located at the chromosomal regions 3p21, 2p22, and 2p16, respectively. Of note, chromosome 3 is frequently deleted in choroidal melanomas.4
Our group and others have reported the presence of MSI-L, reduced MMR protein expression, and MMR gene mutations in the cutaneous melanocytic lesions.5,6 These studies support a role for these genetic changes in the pathogenesis of the melanocytic lesions. Notably, similar reports regarding the ocular counterparts are thus far absent. We hypothesized that choroidal melanoma tumorigenesis involves comparable alterations of the microsatellite repeats and MMR proteins. To test our hypothesis, we used polymerase chain reaction–based microsatellite assays and immunoperoxidase staining methods to analyze 57 cases of choroidal melanomas. We addressed 4 questions: is there MSI in choroidal melanomas, and if so, what is the pattern? Are there any alterations in MMR protein expression in these lesions? Is there a relationship between MSI and MMR expression? Is there a relationship between MSI and the clinicopathological features of choroidal melanomas?
Formalin-fixed, paraffin-embedded tissues of 57 choroidal melanomas were obtained from the archival files of the Collaborative Ocular Melanoma Study, 1988 to 2000. The clinical data and the tumor characteristics are shown in Table 1. Histopathological classification and staging were performed according to Armed Forces Institute of Pathology (Washington, DC) modification of the Callender classification.1 Sections were stained with hematoxylin-eosin to identify the areas of interest, and stereomicroscopy was used to microdissect additional sections, 5 μm thick, by the Turbett method.7 These sections, from matched tumorous and nontumorous tissues, were stained with eosin. The nontumorous tissues included adjacent retina, choroid, and corneal epithelium. We extracted DNA from these tissues using the DNeasy Tissue Kit (Qiagen, Valencia, Calif) as described by the manufacturer.
The 5 chromosomal regions investigated in this study were chosen for several reasons. First, choroidal melanomas have reasonably consistent karyotypic abnormalities at the 1p36-32 and 9p22 regions.8- 14 Second, the 17p and 9p regions harbor tumor suppressor genes with their protein products (p16, p15, and apoptosis-related factor) being deregulated in these lesions.15- 17 Third, our previous studies established the high sensitivity of these loci at these regions in the detection of MSI in the melanocytic lesions.4- 66 Finally, although genes at these regions are rarely mutated in choroidal melanomas, the main purpose of studying these loci was not for the genes located there but simply to establish the microsatellite status, ie, whether they are microsatellite stable, MSI-L, or MSI-H.
The 11 markers investigated in this study were chosen for the following reasons. First, the markers at 1p36-32 regions (D1S2740, D1S2734, D1S233, D1S2832, D1S513, and MYCL1) were tested previously and proved to be reasonably sensitive in the detection of MSI in the melanocytic lesions. Interferon α and TP53 reside near critical p16 and p53 tumor suppressor genes involved in the pathogenesis of these tumors.15- 17 Three additional markers (BAT40, BAT26, and BAT25) at the 1p, 2p, and 4q regions were used to further define the type of instability (MSI-L vs MSI-H). All the markers were obtained from Research Genetics (Huntsville, Ala), and their characteristics are shown in Table 2.
Polymerase chain reaction was performed in a 10-μL volume containing 1× polymerase chain reaction buffer (10 mmol/L Tris, 50 mmol/L KCl [pH 8.3], 0.02% Tween 20); 0.2 mmol/L each of unlabeled and labeled ATP primers (γ-33P; ICN Biomedicals, Costa Mesa, Calif); 1 μL of DNA supernatant; 0.2 U of Taq DNA polymerase (PGC Scientifics, Frederick, Md); and 125 mmol/L each of dNTPs and 1.5 mmol/L MgCl2. Polymerase chain reaction was performed using a “touch-down” approach: 5 minutes at 94°C; 36 to 38 cycles of 1 minute at 94°C, 1 minute at 60°C to 56°C and 1 minute at 72°C; and 9 minutes at 72°C for final elongation. Hot-start was also employed by adding Taq DNA polymerase when the temperature in the initial ramp was greater than 80°C. Polymerase chain reaction products were resolved on a 6% Long Ranger sequencing gel (FMC, Rockland, Md) and exposed to BioMax MR film (Kodak, Rochester, NY) for 24 to 72 hours. All polymerase chain reaction amplifications and gel loadings were repeated at least twice, and only consistent results were reported.6
Microsatellite instability was defined as variations in the banding pattern between tumor and matching normal DNA. Lesions with MSI at 1 or more loci were scored as MSI positive. Instability at more than 30% of the tested loci was used as a cutoff between MSI-L and MSI-H. Lesions without variations in the banding pattern were labeled as microsatellite stable.19
Immunostaining was carried out as described in previous studies.20,21 Briefly, sections mounted on glass slides were deparaffinized and rehydrated through graded alcohols to water. Endogenous peroxidase activity was blocked with 0.6% hydrogen peroxide in methanol. Sections were then immersed in the retrieval solution (10 mmol/L sodium citrate buffer, pH 6.0) and subjected to heat-induced antigen retrieval for 20 minutes. The slides, in plastic Coplin jars containing retrieval solution, were microwaved at high ( ~ 750 W) for 4 cycles of 5 minutes’ duration each. Nonspecific protein binding was blocked with 10-minute exposure to 10% normal goat serum. Sections were then incubated with mouse monoclonal antibodies as primary antibodies (clone G168-15 for hMLH1 and clone FE11 for hMSH2, both purchased from Oncogene Science, Cambridge, Mass, at a dilution of 1:500) for 30 minutes at room temperature. After brief rinsing in phosphate buffer solution, a catalyzed signal amplification system (K1500 from DAKO Corp, Carpinteria, Calif) was used according to the manufacturer’s instructions. Sections were next treated with peroxidase-labeled streptavidin for 30 minutes at room temperature and incubated with 1,4-diaminobenzidine and 0.06% hydrogen peroxide for 5 minutes. They were counterstained with hematoxylin, dehydrated, cleared, and mounted under coverslips.
Staining intensity, percentage of positive tumor cells (PP), and immunoreactivity score were evaluated as described in previous studies.6 Briefly, staining intensity was scored according to the expression levels in the positive controls as strong, moderate, weak, or negative. The PP was assessed by counting at least 5 different areas at a magnification of ×400. The resulting immunoreactivity score for tumors was determined by multiplying staining intensity by PP; results were scored as negative (0-1), weak (2-3), moderate (4-6), or strong (8-12).
As previously described, additional sections of the samples were stained in parallel but with omission of the primary antibody.6 The retinal cells (the inner and outer nuclear layers) and corneal epithelium served as positive controls.
Statistical analysis was done using the Fisher exact test (Statistix for Windows; Analytical Software, Tallahassee, Fla), analysis of variance, and Spearman correlation coefficient tests. Differences were considered statistically significant at P<.05.
Focal MSI (alterations at some but not all the examined loci) was found in these lesions. The overall prevalence (ie, number of tumors with MSI/total number of the lesions) was 35% (20/57). The highest MSI rates were found at the 1p region, followed by the 17p and 9p regions, respectively (90% [18/20] vs 15% [3/20] vs 10% [2/20]). These results, however, could be related to the use of more markers at the 1p region (Table 2 and Table 3).
The frequency of MSI (number of unstable markers/total number of markers) ranged from 9% (1/11 markers) to 27% (3/11 markers); ie, there was MSI-L. No instabilities were found at the 2p and 4q chromosomal regions. The level of MSI among different loci ranged from no change (BAT25, BAT26, and BAT40) to frequent alterations (D1S2734 and D1S2832). The highest level was observed with D1S2734 and was highest for markers with dinucleotide repeats, lower with tetranucleotides (MYCL1), and absent with mononucleotides (BAT25, BAT26, and BAT40) (Figure 1).
The appearance of additional bands (MSI-1 pattern) and band shifting (MSI-2 pattern) were found in 75% and 25% of tumors, respectively, as compared with autologous normal tissue6(Figure 2). The retinal cells (the inner and outer nuclear layers) and corneal epithelium demonstrated strong nuclear staining for the repair proteins (Figure 3). As shown in Table 4, the average values of PP, staining intensity, and immunoreactivity score were insignificantly reduced in the unstable tumors when compared with the stable ones (for example, PP showed P = .62 and P = .74 for hMLH1 and hMSH2, respectively).
The association between MSI and selected clinicopathological features of choroidal melanomas was examined and tested for statistical significance. There was a trend (P<.10) toward earlier age at onset and tumor cells in vascular lakes among the MSI-L subgroup. Mitotic activity; ciliary body involvement; invasion of the blood vessels, long ciliary nerve, vortex veins, and scleral lamellae; and extraocular extension were insignificantly higher among the unstable tumors than the stable ones. There was no significant difference in survival between patients with MSI-L and microsatellite-stable tumors (Table 1 and Figure 4).
To date, little has been determined about MSI and MMR protein status in choroidal melanomas. Moreover, little is known about the association of these alterations with the clinicopathological characteristics of these tumors. We carried out the present investigation to gain insights into these issues. Our data clearly demonstrate that the MSI-L pattern defines a subset of choroidal melanomas. In contrast, using a panel of microsatellite markers roughly half the size of ours, one prior study reported MSI in only 1 of 52 uveal melanomas.22
The observation of MSI-L in choroidal melanomas agrees with other studies in their cutaneous counterpart5,6 and the notion that tumors outside human nonpolyposis colonic carcinoma have MSI-L rather than MSI-H patterns.5,6 It also suggests the involvement of MSI in choroidal melanoma tumorigenesis. Four different possibilities may explain the MSI-L pattern in the choroidal melanomas. First, the variable expression of MMR genes with weakly penetrant mutations and attenuated phenotype may be manifested as MSI-L. A similar finding has been observed in yeast, where inactivation of hMSH3 results in a less severe phenotype than inactivation of hMSH2.23 Second, there may be inherent intrinsic instability of these loci. This is an unlikely alternative because none of the tumors derived from other tissues, such as colorectal adenomas and hepatocellular carcinomas,24,25 demonstrated MSI at any of these loci. Third, there may be inactivation of non-MMR genes or additional MMR genes other than those encountered in human nonpolyposis colonic carcinoma, such as the hMSH6 gene.26 By far, the most important loci involved in choroidal melanoma map to chromosomes 3, 6, and 8. Therefore, future studies are mandated to examine their microsatellite status. Fourth, it is still possible that MSI also reflects an increased rate of spontaneous mutations in these lesions.
The 2 MSI banding patterns observed in our study are similar to those described in human nonpolyposis colonic carcinoma, familial ovarian carcinomas, and melanocytic skin tumors.5,6 The MSI-1 pattern, entailing the presence of additional bands, may be due to division of the sequence within 1 of the original alleles.27,28 The band-shifting pattern (MSI-2) may be due to the migration of CA-repeat bands as a result of instability of those repeats at 1 or both alleles.27 To the best of our knowledge, the present investigation is the first to characterize the banding pattern in the choroidal melanomas.
To investigate the underlying reasons for this MSI-L pattern, we further examined the MMR protein expression and its correlation with MSI status in choroidal melanomas. The presence of strong nuclear MMR protein expression, particularly in the retina and corneal epithelium, suggests transcriptional and translational control analogous to that of other proteins involved in DNA replication and compartmentalization of the MMR proteins and the presence of a nuclear localization signal. It is also consistent with the MMR protein biochemical function in DNA repair.29,30
In agreement with similar findings in their cutaneous counterparts, MMR protein expression was less in choroidal melanomas with MSI-L as compared with the microsatellite-stable lesions. However, this reduction was not statistically significant and suggests (1) the majority of the tumor cells in these lesions may carry functional MMR genes, (2) these tumors may harbor missense mutations in the MMR genes that result in dysfunctional but still detectable proteins, or (3) these lesions could have a defect in other unknown MMR or non-MMR genes or epigenetic mechanisms.26,31 As MMR genes fit into the Knudsen 2-hit theory of tumor suppressor genes, a germline mutation in 1 allele cannot abolish the repair function. An additional somatic mutation in the second allele would be required for complete loss of expression and for the development of MSI.18,32
From the diagnostic point of view, the presence of MSI-L has been used as an adjunctive diagnostic tool in other tumors.33 Our data, along with extended screening and follow-up of a larger sample size in future studies, may help establish a consensus panel of markers for adjunctive diagnostic use in choroidal melanomas. The finding of MSI-L in choroidal melanomas might also have some prognostic value as it has in some other types of tumors.34 Although not statistically significant in the present series of cases, some of the clinicopathological associations we examined did exhibit a trend (P<.10) toward statistical significance. They might prove to be of prognostic utility if substantiated by larger studies in the future.
In our series, there was no significant difference in survival between the patients with stable and MSI-L choroidal melanomas. This finding contrasts with the findings in breast cancer, where MSI was associated with a worse prognosis,35 and with the findings in human nonpolyposis colonic carcinoma, where MSI is associated with better prognosis.24,36,37 These differences could be explained by 2 possibilities: the association between MSI and prognosis has organ specificity and MSI may target different genes in different organs with different impact on the prognostic outcomes.
We report here the analysis of both MSI alterations and MMR protein expression in choroidal melanomas. Our findings suggest that MSI may be involved in the development of at least a subset of these tumors. The fundamental causes and functional effects of MSI alterations in choroidal melanomas are yet to be determined. To the best of our knowledge, this is the first report investigating choroidal melanomas for MSI and the expression of MMR proteins.
Correspondence: Gary S. Wood, MD, Department of Dermatology, University of Wisconsin, One South Park, Seventh Floor, Madison, WI 53715 (firstname.lastname@example.org).
Submitted for Publication: August 4, 2004; final revision received December 13, 2004; accepted December 19, 2004.
Financial Disclosure: None.
Funding/Support: This work was supported by merit review funding from the Department of Veteran Affairs, Washington, DC, and grant AR02136 from the National Institutes of Health, Bethesda, Md.