Ocular melanoma may be more prevalent among patients with light irises than those with dark irises.
To examine a large clinical series of patients with intraocular melanoma to determine if light irises are associated with increased risk of death from these tumors.
A total of 1162 patients treated with proton irradiation between 1984 and 1996 were observed through 1997.
Iris color in the patients was blue or gray in 48%, green or hazel in 30%, and brown in 23%. Tumors in patients with blue or gray irises were less heavily pigmented (P<.001) and closer to the optic disc and macula (P<.001). Five- and 10-year metastasis-related death rates were 0.14 and 0.21, respectively, for those with blue or gray irises and 0.10 and 0.15, respectively, for those with darker irises (P=.02). In a Cox proportional hazards regression controlling for tumor characteristics, patients with blue or gray irises died of metastatic disease at a rate 1.90 times (95% confidence interval, 1.26-2.85) that of patients with brown irises. The rate of metastatic death was not significantly elevated for those with green or hazel irises (relative risk, 1.43; 95% confidence interval, 0.91-2.23).
Patients with blue or gray irises appear to be at increased risk of metastatic death from choroidal melanoma, independent of other risk factors.
OCULAR MELANOMA may be more prevalent among persons with light irises than those with dark irises. In 3 case-control studies1-3 investigating the role of sunlight in the etiology of melanoma of the uveal tract, those with the disease were more likely than controls to have blue or gray irises. An association has also been demonstrated between light iris color and the presence of melanocytic lesions of the iris.4,5 Based on a retrospective analysis of data from a large clinical series of patients with intraocular melanoma, we report for the first time, to our knowledge, an association of iris color with risk of death from these tumors.
The subjects comprise a consecutive series of 1653 patients treated with proton beam irradiation for intraocular melanoma between 1984 and 1996. Patients were included in the study if they had unilateral disease and an untreated tumor involving the choroid or ciliary body, showed no evidence of metastatic disease on pretreatment examination, were US citizens, were treated with the standard therapeutic dose of 70 cGy, and did not receive adjuvant chemotherapy. Pretreatment screening for metastasis included a physical examination and chest x-ray examination. The patients were treated at the Harvard Cyclotron Laboratory, Cambridge, Mass, as previously described6 and followed up through April 1997. Survival status was updated annually. Nine patients had not been contacted for at least 24 months, and recent follow-up information was available for the remainder. In 57 cases, the anterior margin of the tumor extended to the iris. Iris color was assessed at baseline and recorded as blue or gray, green or hazel, or brown. Tumor pigmentation was assessed during preirradiation surgery to suture markers around the tumor and recorded as none, minimal, moderate, or heavy. Iris color and tumor pigmentation were recorded for 1162 of the 1287 eligible cases. Analyses were restricted to these patients.
The Kaplan-Meier method7 was used to estimate survival rates. Univariate comparisons were made using the Student t, rank sum, and χ2 tests. Cox proportional hazards analysis8 was used to estimate rate ratios and survivor functions for iris color adjusted for other prognostic factors.
Iris color in the patients was blue or gray in 48%, green or hazel in 30%, and brown in 23%. Median follow-up time (since treatment) among survivors was 75 months. There were 185 deaths due to metastasis and 143 due to other causes as documented by medical records and pathology reports (81%), death certificates (11%), or next-of-kin account (9%).
Patient and tumor characteristics are given by iris color in Table 1. Patients with blue or gray irises did not differ from those with darker irises in age at treatment, sex, laterality, largest tumor diameter, tumor height, or anterior extent. Heavy tumor pigmentation was more common among patients with dark irises than those with blue or gray irises. Tumors were closer to the optic disc and macula in patients with blue or gray irises than in those with darker irises.
Survivor functions by iris color were estimated for death due to all causes and for metastasis-related deaths, censoring deaths due to other causes at the date of death. Survival was poorer among those with blue or gray irises (P=.02). Five- and 10-year rates of tumor-related death were 0.14 and 0.21, respectively, for those with blue or gray irises and 0.10 and 0.15, respectively, for those with darker irises. The functions for all-cause mortality did not differ significantly by iris color (P=.07), with 5- and 10-year rates of 0.21 and 0.35, respectively, for blue or gray irises, and 0.17 and 0.29, respectively, for darker irises. Separate survivor functions for blue or gray and darker irises, adjusted for tumor pigmentation and proximity to the macula, were derived using the Cox regression model (Figure 1). Survival was poorer among those with blue or gray irises (P<.001).
We performed a Cox proportional hazards regression including terms for tumor pigmentation, diameter, height, anterior margin, distance from the macula, and iris color. Patients with blue or gray irises died of metastatic disease at a rate 1.90 times that of patients with brown irises (Table 2). The rate of metastatic death was not significantly elevated for those with green or hazel irises. Elevated risks were associated with heavy tumor pigmentation (P<.001) and large tumor diameter (P<.001 for tumors >15 mm).
We further examined the joint effects of tumor pigmentation and iris color on tumor survival, contrasting high- and low-risk levels of these factors (Table 3). Tumor death rates in patients with heavy tumor pigment and dark irises, or light tumor pigment and light irises, were about twice that of patients at low risk on both factors (relative risk, 2.1 and 1.8, respectively). Heavy tumor pigment and light iris color was associated with more than a tripling in death rates. However, there was no evidence of an interaction between the 2 factors (P=.84), producing higher (or lower) than expected rate ratios when both were present.
The results of this study suggest that patients with blue or gray irises are at increased risk of metastatic death from choroidal melanoma, independent of other risk factors. Risk associated with green or hazel iris color may also be elevated to a lesser degree. To our knowledge, this represents the first report of a prognostic influence for iris color in progression and death from these tumors.
As observed in previous studies, heavy tumor pigmentation was independently associated with an elevated risk of death from metastatic disease. Tumor pigment was positively associated with degree of pigment in the iris. We are not aware of any previous study reporting this association. In one study9 comparing black patients enucleated for choroidal and ciliary body melanomas with white control subjects, the tumors in black patients were more heavily pigmented that those of white patients on histopathologic examination. Iris color was not reported in that study, but it may be presumed that light irises were more common among the white patients than the black patients. It is possible that the apparent association between pigmentation of the tumor and iris results from method of assessment used: clinical assessment of tumor pigment might be influenced by the degree of pigment in the overlying retinal pigmented epithelium, a characteristic that may be associated with iris pigmentation. We reviewed a series of 97 pathology reports documenting iris color and tumor pigmentation to see if the association between iris color and tumor pigmentation noted in the present series held when the tumors were assessed histologically. The proportion of blue or green eyes (79%) was similar to that observed in the present series (77%). As in this clinical series, tumors graded as heavily pigmented were less likely to be associated with light iris color (17% vs 40% in blue or green vs brown eyes, P=.03).
It is accepted that sunlight exposure is an important risk factor for cutaneous melanoma.10 This disease is more prevalent among fair-skinned people whose relative lack of skin pigmentation renders them more susceptible to damage from UV radiation. Three1-3 of 41-3,11 case-control studies have demonstrated an excess risk of ocular melanoma among those with light irises, which the authors hypothesized may be due to sunlight exposure early in life2 or to large doses of sunlight received intermittently during recreational activities.1 One study10 reported a positive association between some measures of sunlight exposure and risk of ocular melanoma despite finding no risk elevation attributable to light iris color. A connection between sunlight exposure and ocular melanoma in those with blue irises is consistent with our finding that the tumors of patients with blue irises were significantly closer to the optic disc and macula, the region of the choroid most directly exposed to sunlight. More research is needed to firmly establish this link.12
Because the patients in our series received eye-conserving therapy, we are unable to investigate whether histopathologic differences may be associated with iris color. A study of enucleated patients is needed to determine whether differences in cell type between those with light and dark irises explain the poorer survival experienced by patients with light irises.
Results from this study suggest that iris pigment has an influence not only on the risk of developing choroidal melanoma but also on the probability of surviving the disease. How iris color might play a role in the metastatic spread of these tumors is unclear. Although iris and tumor pigment are correlated, the mechanisms by which these factors influence prognosis appear to be distinct: the combined effect of light iris color and dark tumor pigment was consistent with a simple additive relation (Table 3). Tumors in eyes with light irises were more likely to be positioned centrally in the posterior pole (ie, with a margin near the optic disc and macula) where light exposure is most direct. Our findings raise the interesting possibility that specific genetic lesions brought about by UV light exposure in the eye promote ocular melanoma and are associated with a more aggressive form of the disease.
Accepted for publication January 8, 1999.
Dr Gragoudas is a senior scientific investigator for Research to Prevent Blindness Inc, New York, NY.
Reprints: Evangelos S. Gragoudas, MD, Retina Service, Massachusetts Eye and Ear Infirmary, 241 Charles St, Boston, MA 02114.
et al. Risk factors for ocular melanoma: Western Canada Melanoma Study. J Natl Cancer Inst.
1985;74775- 778Google Scholar
JF Sunlight exposure as a risk factor for intraocular malignant melanoma. N Engl J Med.
1985;313789- 792Google ScholarCrossref
DK Uveal melanoma in relation to ultraviolet light exposure and host factors. Cancer Res.
1990;505773- 5777Google Scholar
RP Color as a risk factor in iris melanoma. Am J Ophthalmol.
1984;98558- 561Google Scholar
JA Association between iris color and iris melanocytic lesions. Am J Ophthalmol.
1985;100547- 548Google Scholar
JM Charged particle irradiation of uveal melanoma. Schachat
St Louis, Mo Mosby–Year Book Inc1994;785- 794Google Scholar
P Nonparametric estimation from incomplete observations. J Am Stat Assoc.
1958;53457- 481Google ScholarCrossref
D Analysis of Survival Data. London, England Chapman & Hall1984;
IW Malignant melanoma of the choroid and ciliary body in black patients. Arch Ophthalmol.
1984;10277- 79Google ScholarCrossref
DS Malignant melanoma: perspectives on incidence and its effect on awareness, diagnosis and treatment. CA Cancer J Clin.
1996;46195- 198Google ScholarCrossref
PH Host factors, UV radiation, and risk of uveal melanoma. Arch Ophthalmol.
1990;1081274- 1280Google ScholarCrossref
DM Epidemiologic aspects of uveal melanoma. Surv Ophthalmol.
1988;32239- 251Google ScholarCrossref