If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Pathology, Leiden University Medical Center, Leiden, The NetherlandsDepartment of Pathology, Section Ophthalmic Pathology, Erasmus MC University Medical Center, Rotterdam, The Netherlands
Heavy pigmentation is known to be a prognostic risk factor in uveal melanoma (UM). We analysed whether genetic tumour parameters were associated with tumour pigmentation and whether pigmentation should be included in prognostic tests.
Retrospective comparison of clinical, histopathological and genetic features and survival in UM with different pigmentation.
1058 UM patients from a white European population with diverse eye colours, enucleated between 1972 and 2021.
Cox regression and log-rank test were used for survival analysis, chi-square test and Mann-Whitney U test were used for correlation analysis.
Main Outcome Measures
UM-related survival based on tumour pigmentation and chromosome status, correlation of tumour pigmentation with prognostic factors.
5-year UM-related mortality was 8% in patients with non-pigmented tumours (n=54), 25% with lightly-pigmented tumours (n=489), 41% with moderately-pigmented tumours (n=333) and 33% with dark tumours (n=178) (p<0.001). The percentage of tumours with monosomy 3 or 8q gain increased with increasing pigmentation (31%, 46%, 62%, and 70% having monosomy 3, p<0.001, and 19%, 43%, 61%, and 63% having 8q gain, p<0.001, in the four increasing pigment groups, respectively). BAP1 loss (known for 204 cases) was associated with an increased tumour pigmentation (p<0.001). Cox regression analysis on survival showed that when chromosome status and pigmentation were both included, pigmentation was not an independent prognostic indicator. PRAME expression was a significant prognostic marker in light (p=0.02) but not in dark tumours (p=0.85).
Patients with moderately and heavily-pigmented tumours showed a significantly higher UM-related mortality than patients with unpigmented and light tumours (p<0.001), supporting prior reports on the relation between increased tumour pigmentation and a worse prognosis. While we previously showed that a dark eye colour was associated with tumour pigmentation, we now show that also the tumour’s genetic status (chromosome 3 and 8q/BAP1 status) is related to tumour pigmentation. When pigmentation and chromosome 3 status are both included in a Cox regression analysis, pigmentation is not an independent prognostic factor. However, evidence from this and previous studies shows that chromosome changes and PRAME expression have a stronger association with survival when they occur in light tumours than in dark ones.
. UM originates from melanocytes in the uvea, which are neural crest-derived cells that contain melanosomes and give rise to pigmentation. UM vary in their level of pigmentation, both macroscopically and microscopically, from completely amelanotic to heavily pigmented; some tumours show areas with different degrees of pigmentation, which can even be visualized by MRI (magnetic resonance imaging)
The degree of tumour pigmentation is part of the standard histopathological analyses of UM and several reports show an association between heavy pigmentation and a poor prognosis. McLean et al. studied small choroidal and ciliary body melanomas and reported that a high degree of microscopic pigmentation constituted a negative prognostic factor; they suggested that this prognostic significance was related to the higher number of epithelioid cells in dark tumours
. Seddon et al. showed that tumour pigmentation was one of the five factors that best predicted prognosis (the other four being the presence of epithelioid cells, tumour diameter, location of the anterior tumour margin and invasion of the transection line)
. Similarly, the Collaborative Ocular Melanoma Study (COMS) reported that UM with heavy microscopic pigmentation tended to be larger and contained more epithelioid cells than tumours with little pigmentation
. Regan et al. studied eyes with UM treated with proton beam radiation and showed that both blue eye colour and heavy clinical pigmentation were associated with a worse survival, and that the risk of death was even higher in patients with blue eyes and dark tumours
. Markiewicz et al. analysed macroscopic (clinical) pigmentation as well, and compared 26 amelanotic and 128 pigmented UM. Their findings agree with previous studies in terms of survival, with the added report of a higher proportion of BAP1 negative cases among the pigmented tumours.
Several of these reports mention that increased pigmentation was associated with the presence of epithelioid cells, which is known to be associated with chromosome 3 loss, one of the strongest risk factors for developing metastases
. Furthermore, monosomy 3 had a greater influence on survival in patients with light eyes than in those with dark eyes.
We wondered whether tumour pigmentation is an independent prognostic factor for survival in patients with a UM and whether tumour pigmentation is not only associated to the genetically-determined iris colour but also related to the tumour’s genetic status. Therefore, we set out to compare histopathologic tumour characteristics, chromosome 3 and 8q status and BAP1 expression in tumours with different pigmentation levels. We also looked at the relation between these genetic features and survival in UM patients with different degrees of histological tumour pigmentation.
This project adhered to Dutch law and the tenets of the Declaration of Helsinki (World Medical Association of Declaration 1964; ethical principles for medical research involving human subjects) Materials and histopathological, genetic and follow-up data are part of Biobank OOG-2 of the Leiden University Medical Center (LUMC) (Uveamelanoomlab-2019-7, approval May 2019). Permission was given to use these data for this analysis by the Medical Ethics Committee (no.: B20.022).
We performed a retrospective cohort study of UM patients enucleated at the Leiden University Medical Center between August 1972 and October 2021 and selected 1197 cases in which tumour pigmentation had been recorded. We excluded patients who had received some form of irradiation treatment before enucleation. The total number of analysed cases was 1058. Tumour pigmentation was scored macroscopically by the operator during dissection of the eye in two halves, as per standard preparation for pathological analysis. The pigmentation was scored on a 4-point scale as follows: 1 – unpigmented (white), 2 – lightly pigmented (grey), 3 – moderately pigmented (brown), 4 – heavily pigmented (dark brown – black). Examples of tumours in each pigmentation group can be found in Supplemental Figure 1. In case of tumours with heterogenous pigmentation, the pigmentation of the most prominent component was scored. When only two groups of tumour pigmentation were used for analysis, groups 1 and 2 were classified as light and groups 3 and 4 were classified as dark. Cell type, ciliary body involvement, episcleral growth, tumour diameter and thickness were evaluated during the histopathological examination performed by one of three experienced pathologists. The AJCC stage was scored according to the 8th edition of the AJCC staging manual
. Chromosome copy number was determined either by karyotyping, fluorescence in situ hybridization or SNP array, and when any of these tests showed an abnormality, the case was considered aberrant (monosomy 3 or chromosome 8q gain)
and scored by an ophthalmic pathologist. Clinical and survival information was collected from patient charts, follow-up time was defined as the time from enucleation to death or last recorded patient contact. Information on death and cause of death was obtained from the Integral Cancer Center West. The median follow-up of this population was 57 months (range 0 – 547 months, mean 110).
Eye colour information was available for 412 cases and it was collected from medical charts, clinical photographs and self-reported iris colour as previously described
. Iris colour was scored as blue/grey, green/hazel and dark brown.
PRAME mRNA expression
PRAME expression was available for 64 cases. mRNA expression was measured from archived snap-frozen material on an Illumina HT-12v4 chip (Illumina) using probe ILMN_1700031. PRAME expression was classified as positive or negative according to a cut-off value set at the inflection point of the PRAME expression curve obtained using probe ILMN_1700031 19.
Statistical analyses were performed with the statistical software SPSS (IBM SPSS Statistics for Windows, Version 25.0; IBM Corp). Survival was evaluated with Kaplan-Meier curve and log-rank test, and with Cox regression, correcting for age and sex. In both cases, patients who died with UM metastases were considered as events and patients alive at the end of follow up, lost to follow-up or dead due to other causes were censored. Categorical variables were compared with Pearson’s chi-square test, while continuous variables were compared with Mann-Whitney U test or Kruskal-Wallis test, in analyses with more than two groups. Logistic regression was performed with tumour pigmentation as a binary dependent variable (light or dark) and eye colour and chromosome 3 status as categorical independent variables. A p value < 0.05 was considered significant. In addition, Bonferroni correction was applied in Table 4 and Supplemental Table 7 and 8 and the adjusted α is reported in the footnote of each table. Because of the strong association between variables tested, we deemed a multivariable analysis not to be an ideal method to reduce multiple testing bias in this case.
Table 4Comparison between clinical and histopathological and genetic features and tumour pigmentation (4 groups) in 1058 UM patients
Low pigmentation (489)d
Moderate pigmentation (336)d
Heavy pigmentation (178)d
Age at enucleation (years)e
Median Follow up (months)e
Largest Basal Diametere
Epithelioid or mixed (700)
Ciliary body involvement
Chromosome 3 status
BAP1 positive (79)
BAP1 negative (125)
a: Pearson’s χ2 test
b: Kruskal-Wallis test
c: Percentages are rounded and may not total 100
d: Percentages were calculated excluding missing data
The Leiden UM enucleation database includes all UM cases that have been enucleated for UM at the LUMC (Leiden University Medical Center, Leiden, The Netherlands) from 1972 to 2021. Of these 1291 UM cases, information regarding macroscopic tumour pigmentation at the time of enucleation was available for 1197 cases. Since irradiation might influence tumour features, we excluded UM that had undergone treatment such as radio-active plaque irradiation or proton beam therapy prior to enucleation. This left 1058 cases for our study.
Based on macroscopic analysis of enucleated UM, tumours were assigned to one of four pigmentation categories. Of the 1058 cases in our cohort, 55 (5%) were classified as unpigmented, 489 (46%) as lightly pigmented, 336 (32%) as moderately pigmented and 178 (17%) as heavily pigmented. We first set out to compare pigmentation with UM-related mortality to determine whether the level of pigmentation had prognostic value in our European patient group. Four cases did not have follow-up data and were excluded from survival analyses.
Patients with moderately and heavily-pigmented tumours showed a significantly worse UM-related overall survival than patients with unpigmented and light tumours (Figure 2, Table 1a) (p < 0.001, both for Kaplan-Meier curves and Cox regression). Supplemental Table 2 shows the results of Cox regression analysis comparing dark and light tumours (p < 0.001, Supplemental Table 2). The difference in survival between the moderate and heavy pigmentation groups was not significant (Cox regression: p = 0.09, Supplemental Table 3).
Table 1Tumour pigmentation and chromosome 3 and 8q status versus UM-related survival in 1054 UM patients., Table 1a. Cox regression for effect of macroscopic histological pigmentation (4 groups) on UM-related survival, correcting for age and sex in 1054 UM patients.
We then compared the clinical, histopathological, and genetic characteristics of the four pigmentation groups. As shown in Table 4, an increase in pigmentation correlated with an older age at diagnosis (p < 0.001), a higher frequency of occurrence of an epithelioid and mixed cell type (p = 0.001), more frequent ciliary body involvement (p < 0.001), a higher AJCC (American Joint Committee on Cancer) stage (p = 0.001), and shorter median follow-up (p < 0.001). These findings show that our data agree with prior results, i.e. that dark tumours more often show clinical and histopathological features associated with a poor prognosis. We subsequently compared genetic features: more pigment was associated with a higher proportion of tumours carrying monosomy 3 (M3) compared to disomy 3 (D3) (p < 0.001), chromosome 8q gain compared to normal 8q status (p < 0.001), and loss of BAP1 staining (p = 0.001) (Table 4). Therefore, we clearly demonstrate that, in addition to having adverse clinical and histopathological features, dark tumours more often show prognostically-unfavourable chromosome aberrations.
Association of chromosome status and eye colour with tumour pigmentation
While we now show a correlation between tumour and chromosome 3/BAP1 status and chromosome 8q (Table 4), we previously
reported that eye colour is also related to tumour pigmentation: patients with brown eyes more frequently had dark tumours compared to patients with blue or green eyes, who more often had light tumours. These findings confirm a previous report by Regan et al.
. In order to understand whether eye colour and chromosome 3 and 8q status are independently associated to tumour pigmentation, we computed a logistic regression with tumour pigmentation as the binary dependent variable, and eye colour and chromosome 3 and 8q status as the independent variables. As Table 5 shows, both M3 and brown eye colour were significant predictors of tumour pigmentation (D3 with 8q gain: p 0.009, OR = 3.138, C.I. 1.337-7.364, Wald = 6.902, M3: p < 0.001, OR = 3.229, C.I. 1.855-5.620, Wald = 17.189, brown eye colour: p = 0.002, OR = 3.271, C.I. 1.524-7.016, Wald = 9.258).
Table 5Logistic regression testing the predictive power of eye colour and chromosome 3 and 8q on the level of tumour pigmentation (light vs dark)†
D3, no 8q gain*
D3, 8q gain
†: light = unpigmented + low pigmentation; dark = moderate pigmentation + heavy pigmentation
The chi-square test in Table 6 shows that M3 and 8q gain were especially related to increased tumour pigmentation in the blue eye group (p = 0.001), with a lower but still significant association in the brown eye group (p = 0.04) (Table 6).
Table 6Distribution of chromosome 3 status and chromosome 8q status according to tumour pigmentation in 3 groups with different eye colours (p value determined by Chi square test). Eye colour groups are analysed separately and percentages are calculated per eye colour group.
Is pigmentation a prognostic factor independent of chromosome 3 status?
To determine whether pigmentation was related to survival independently of chromosome 3 and 8q status, we used a Cox regression model (Table 1b) which included both eye colour and chromosome 3 and 8q status, while correcting for age and sex. Once chromosome 3 and 8q status were taken into account, tumour pigmentation did show a significant association with survival.
Table 1bCox regression for effect of macroscopic histological pigmentation (2 groups) and chromosome 3 and 8q status on UM-related survival, correcting for age and sex in 1054 UM patients.
Dark vs light Pigmentation‡†
Monosomy 3 vs disomy 3‡
8q gain vs normal 8q
‡: adjusted for age and sex
†: light = unpigmented + low pigmentation; dark = moderate pigmentation + heavy pigmentation
This conclusion can also be reached by analysing the relation between pigmentation and survival in patients with either D3 or an M3 tumours separately (Figure 3). Within the D3 population, the curves for dark and light tumours diverge slightly at 5 years, although the difference does not reach significance (p = 0.13), while in the M3 group, the curves for light and dark tumours never diverge (p = 0.63).
We then compared the distribution of clinical and histopathological tumour features between light and dark tumours in the D3 and the M3 sub-cohorts separately. Among the D3 tumours, dark UM more frequently showed chromosome 8q gain (p = 0.003) (Supplemental Table 7) compared to light UM, whereas in the M3 population, none of the factors analysed showed a significant difference between light and dark UM (Supplemental Table 8).
Influence of pigmentation on impact of prognostic factors
We previously reported that the degree of pigmentation of the tumour influenced the relationship between chromosome 3 and survival: loss of one chromosome 3 or gain of copies of 8q had a much larger effect on survival in patients with a light tumour than in those with a heavily pigmented tumour
, observed an association between heavy pigmentation and poor survival (Table 1, Figure 2). A higher degree of pigmentation was related to larger tumour size, more frequent involvement of the ciliary body and a more frequent occurrence of epithelioid cells, in agreement with prior publications (Table 4). Furthermore, our results showed that darker tumours more frequently have loss of chromosome 3/BAP1 and gain of chromosome 8q (Table 4). The analyses performed in the D3 and M3 cohorts separately showed that chromosome 3 status is the most relevant factor related to prognosis (Figure 3), while in the D3 sub-cohort, dark tumour pigmentation shows an association with chromosome 8q gain (Supplemental Table 7).
The link between chromosome 3 and tumour pigmentation is further supported by the logistic regression model presented in Table 5, which shows that both monosomy 3 and brown eye colour are significant predictors of dark tumour pigmentation.
We wondered why and how monosomy 3 and tumour pigmentation could be related, and we postulate that a possible connection might be the presence of inflammation. Previous findings show that an inflammatory phenotype does not inhibit tumour growth as in other types of cancer, but instead, seems to be associated with the development of metastases
. The number of macrophages was previously shown to be significantly correlated with the degree of tumour pigmentation, both by the COMS study analysing choroidal melanomas and by Mäkitie et al. in choroidal and ciliary body melanomas 7 27. The presence of lymphocytes and macrophages is associated with monosomy of chromosome 3 and carries a bad prognosis
. More recently, a study comparing BAP1-positive and BAP1-negative tumours reported an upregulation of several immuno-suppressive genes in BAP1-negative tumours, thus suggesting that the infiltrate of tumours with BAP1 loss has a regulatory and immunosuppressive T cell phenotype
that the amount of infiltrating macrophages is related both to pigmentation and chromosome 3 status: the group with the least number of macrophages consisted of lightly pigmented tumours with disomy of chromosome 3.
It seems paradoxical that pigmented UM have an increased immune infiltrate but show more UM-related death. It is well possible that specifically the presence of pigment helps to reduce the activity of anti-tumour T cells: in support of an immunosuppressive property of UM cells themselves, Gezgin et al. recently showed that a more successful expansion of UM-reactive Tumour Infiltrating Lymphocytes (TILs) can be achieved when the lymphocytes are separated from the tumour environment early, compared to direct culture of tumour tissue or mononuclear cell enrichment
Evidence from cutaneous melanoma similarly suggests that an increased tumour pigmentation influences tumour behaviour: Brożyna et al reported that patients with amelanotic cutaneous melanomas showed better survival than patients with melanotic melanomas, and that the presence of melanin in melanoma cells decreased the outcome of radiotherapy
. In line with these findings, Slominski et al. reported that pigmented cutaneous melanoma cells were more resistant to cyclophosphamide and to IL2-stimulated peripheral blood lymphocytes when compared to unpigmented cells
. They also showed that the melanogenesis intermediate L-DOPA inhibited proliferation, progression through the cell cycle and function of lymphocytes. These results suggest that melanocytes or pigment-related molecules influence tumour behaviour in cutaneous melanoma, perhaps through an effect on the immune system. While normal cutaneous melanocytes are best-known for their ability to produce melanin and to protect DNA from ultraviolet light damage, they can respond to external signals such as cytokine mediators and hormones and are able to secrete a wide range of cytokines and signalling molecules with neuroendocrine and hormone-like properties
. In addition, the relationship between skin melanocytes and inflammation may be influenced by the level of pigmentation. Tam et al. looked at cytokine production by dark and light melanocytes upon LPS stimulation
. When lightly pigmentated melanocytes were stimulated, they produced more pro-inflammatory cytokines such as CCL2, IL6, and TNF1 compared to dark melanocytes.
These considerations suggest that the presence of inflammation may indeed link chromosome 3 status, inflammation and pigmentation, and that targeting the pigmentation process may help in increasing the susceptibility of melanomas to immunotherapies.
A further element to consider is the link between eye colour and the degree of tumour pigmentation, that was reported by Regan et al and in a previous study from our centre
. Eye colour showed strong association with tumour pigmentation in D3 tumours as well as M3 tumours (Table 6). This finding raises the possibility that tumour pigmentation might be partly influenced by the patient’s genetic background and the intrinsic level of eye pigmentation, which is determined by genes that regulate the amount and the type of melanin present in melanosomes (expressed as the ratio of eumelanin and pheomelanin)
, some of which have shown an association with the risk to develop UM: specific SNPs at rs12913832, rs1129038 and rs916977 in the HERC2/OCA2 locus have been associated with a lower risk of developing UM, while a specific SNP at rs12203592 in the IRF4 locus was associated with a higher risk of UM
. Recently, Mobuchon et al. hypothesised that the patient’s genes that determine eye colour influence which chromosome changes/mutations occur, based on their observation that in a cohort of 972 patients with European ancestry, one of the IRF4 SNPs (rs12203592) was a significant risk factor only in the D3 UM population, while a SNP on the HERC2 locus (rs12913832) was a significant risk factor only in the M3 UM population
. However, when we compare tumour chromosome aberrations in UM from individuals with different eye colours in our Leiden cohort, we do not see differences in the percentages of M3 tumours between different eye colours, with M3 being present in 56% of blue eyes, 47% of green eyes and 57% of brown eyes (p = 0.45)
. This is in contradiction with the findings of Mobuchon. However, an important difference between the Netherlands and France that needs to be considered is that in the Netherlands, the majority of people have blue eyes. In our current study we see that across all three eye colour groups, darker tumours invariably contain a higher proportion of M3 cases than light tumours. Taken together, our data suggest that both chromosome 3 and 8q status and the patient’s genetic pigmentation may cooperate in determining the degree of tumour pigmentation. Further studies are needed to confirm these observations and define the specific role of each of these factors.
A limitation of this study is that we used macroscopic pigmentation, which may be influenced by several factors other than the melanin content of UM cells, such as the presence of necrosis or pigment-laden macrophages. Moreover, since these cases were collected over almost 40 years, there might be inter-observer variability in pigmentation grading. However, the use of a pre-defined 4-point scale developed in 1984 should decrease the variability in subjective evaluation. One should also keep in mind that the clinical observations reported in this manuscript are associations and that further functional studies would be required to determine how monosomy 3 or chromosome 8q gain cause increased melanocyte pigmentation at the cellular level.
We confirm that dark tumours carry a worse prognosis than light tumours and report a significant association between dark tumour pigmentation and negative genetic prognostic factors: monosomy of chromosome 3, loss of BAP1 expression, and gain of chromosome 8q. Tumour pigmentation is not an independent prognostic factor when chromosome status is taken into account, but does influence the prognostic potential of genetic changes in the tumours. Further studies are needed to determine whether the genetical eye colour and therefore the type of melanin (pheo versus eumelanin) in the tumour influences which type of chromosome change/mutation occurs. We also propose that both tumour pigmentation and chromosome 3 status might determine the inflammatory phenotype of UM, hence contributing to the patient’s prognosis. Therefore, respective roles of chromosome 3 status/BAP1 status and pigmentation in the development of an immune response against the tumour should be explored further.
Financial Support: M.C. Gelmi is funded by the Bontius Foundation, Oogfonds, the Sam Fund, the LUF P.A. Jager-van Gelder Fund, the Blinden-Penning foundation and ASROO (Associazione Scientifica Retinoblastoma ed Oncologia Oculare). A.P.A. Wierenga is funded by the European Commission, through the Horizon 2020 grant nr: 667787, UM CURE 2020. The sponsor or funding organisation had no role in the design or conduct of this research.
Tumour pigmentation is associated with poor prognosis in uveal melanoma and is associated with eye colour and chromosome 3 and 8q status.