Background There has been a recent evolution in the management of uveal melanoma in regard to prognostic evaluation and treatment modalities. This study aims to evaluate the epidemiological trends of uveal melanoma since 1973 by using the Surveillance, Epidemiology, and End Results (SEER) Registry.
Methods A total of 7043 uveal melanoma cases from the SEER database were retrospectively analysed, spanning 1973 to 2009. The main outcome measures were disease specific and overall patient survival.
Results The mean patient age at diagnosis was 61.4 years, which has increased over the last 37 years. The mean follow-up interval was 85.9 months. The proportion of tumours demonstrating spindle cell histology has decreased. Spindle cell tumours yielded superior disease-specific survival to epithelioid and mixed tumours. The median age at diagnosis of spindle cell tumours (60 years) was significantly younger than epithelioid tumours (65 years) and mixed tumours (64 years old; p<0.0001). In a Cox proportional hazards regression analysis, tumour histology, T stage and age at diagnosis were associated with disease-specific survival. The prevalence of cutaneous melanoma was increased in patients with uveal melanoma, but the lifetime prevalence of other primary cancers was not appreciably increased.
Conclusions The SEER data set demonstrates epidemiological trends in patient age at diagnosis and tumour histology over the last 37 years. Several patient and tumour characteristics are predictors of disease-specific survival. These findings have implications for disease surveillance and prognostic counselling.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Uveal melanoma confers significant mortality and ocular morbidity. The Collaborative Ocular Melanoma Study (COMS) demonstrated the efficacy of brachytherapy in local control of tumour growth, with survival comparable with enucleation for medium-sized tumours.1 Several centres use external radiation via proton beam irradiation rather than brachytherapy.2–4 Overall, there has been a significant shift away from local tumour resection and enucleation towards vision-sparing radiotherapy in the USA.5 Relative survival rates are estimated at 77–84% at 5 years,5 with 5-year mortality rates estimated to be as high as 53% for large tumours.5 ,6
At present, significant research efforts in uveal melanoma focus on diagnostic modalities, including molecular diagnostics, to provide more specific prognostic information and guide therapy. As such, a greater understanding of trends in epidemiological associations and longitudinal patterns of this cancer may provide additional means to refine prognostic predictions. Singh and colleagues demonstrated stable age-adjusted incidence7 and 5-year relative survival rate5 in uveal melanoma since 1973. The aims of the current study are to evaluate further the epidemiology of ocular melanoma over the past four decades in the USA by using the Surveillance, Epidemiology, and End Results (SEER) Programme Tumour Registry.
Materials and methods
The SEER database is a population-based cancer registry that captures 18 distinct population groups in 198 counties in the USA. It represents approximately 26% of the overall US population and contains information on 7 262 696 cases of cancer diagnosed since 1973.8 All cases of uveal melanoma between 1973 and 2009 were selected from the SEER database, including all 18 registries. Permission to use these data for analysis was obtained from the National Cancer Institute SEER programme. Institutional review board approval was not required for this study.
The data in this study were standardised according to schema published in the second and third editions of the International Classification of Disease for Oncology and the sixth edition of the American Joint Committee on Cancer.9 For the current study, histological type was limited to melanoma (M8720/3, M8721/3, M8722/3, M8726/3, M8730/3, M8742/3, M8743/3, M8761/3, M8770/3, M8771/3, M8772/3, M8773/3 and M8774/3 in the morphological codes of the second edition of the International Classification of Disease for Oncology). Primary site was restricted to uvea, defined for the purposes of melanoma as choroid or ciliary body by site codes C69.3 and C69.4, respectively. Specific details regarding radiation dose, fractionation schedules and adjuvant chemotherapy were not available in the SEER database. Data were analysed using the SEER*Stat Limited Use software provided by the National Cancer Institute, Bethesda, Maryland, and GraphPad Prism software (GraphPad Software, Inc, La Jolla, California, USA). Overall survival and disease-specific survival were the primary outcome measures. Survival curves were generated using the Kaplan–Meier method and compared using the Mantel–Cox log-rank test. One-way analysis of variance was used to compare multiple groups. The Mann–Whitney U test was used to compare continuous variables between two groups. The χ2 test was used to compare SEER cancer rates with published prevalence rates. Cox proportional hazards regression was performed to assess the impact of patient and tumour characteristics on survival. Significance was defined as p<0.05.
The SEER database included 7043 cases of uveal melanoma between 1973 and 2009. Table 1 includes summary statistics for these patients.
The mean patient age was 61.4±15.0 years (range 3–100 years). The cohort was 52.2% men. Cytological, histological or microscopic diagnosis was confirmed on enucleation, biopsy or autopsy specimens in 69.3% of cases. The mean follow-up interval was 85.9±82.3 months. Figure 1 illustrates Kaplan–Meier disease-specific survival for all patients with uveal melanoma.
There were no differences in overall survival (p=0.2182) or disease-specific survival (p=0.9995) when the data were stratified by year of diagnosis. There were no differences in overall survival (p=0.0559; HR=1.070, 95% CI 0.9983 to 1.147) or disease-specific survival (p=0.1503; HR=1.026, 95% CI 0.9281 to 1.134) between men and women. There were also no differences in overall survival (p=0.9131) or disease-specific survival (p=0.3475) among the four main race categories in the SEER database. There was an increase in the mean age at diagnosis for the interval between 1973 and 2009 (figure 2; R2=0.6842).
There were differences in overall survival (p<0.0001) and disease-specific survival (p<0.0001) among histological diagnoses (figure 3). Spindle cell melanoma yielded greater disease-specific survival than either epithelioid melanoma (p<0.0001; HR=0.2947, 95% CI 0.2377 to 0.3653) or mixed epithelioid and spindle cell melanoma (p<0.0001; HR=0.5124, 95% CI 0.4509 to 0.5822). In addition, mixed epithelioid and spindle cell melanoma demonstrated superior disease-specific survival to epithelioid melanoma (p=0.0013; HR=0.7493, 95% CI 0.6285 to 0.8934). The histopathological subtypes of melanoma appeared to vary during the study interval. While the proportion of epithelioid tumours remained relatively constant throughout the study period, there was a decrease in the proportion of spindle cell tumours over the previous four decades (figure 4; R2=0.5352). Also, there was a statistically significant difference in age at diagnosis among the histological subtypes (p<0.001). The median age at diagnosis of patients with spindle cell tumours (60 years old) was significantly younger than with epithelioid tumours (65 years old; p<0.0001) and mixed tumours (64 years old; p<0.0001). However, there was no significant difference in median age at diagnosis between epithelioid and mixed epithelioid and spindle cell melanoma (p=0.3732).
Patients with uveal melanoma were frequently diagnosed with multiple cancers during their lifetimes. Patients diagnosed with uveal melanoma as the first primary cancer fared better than patients with a first non-ocular malignancy in terms of overall survival (figure 5; p<0.0001; HR=0.4812, 95% CI 0.4242 to 0.5459) and disease-specific survival (p=0.0168).
Cox proportional hazards regression analysis was performed for disease-specific survival with the following covariates: gender, histology, T classification (TNM (T, extent of primary tumour; N condition of regional nodes; M, presence or absence of distant metastases) classification), surgery, radiotherapy, first malignancy, race and age at diagnosis. In this multivariate analysis, histology, T classification and age at diagnosis were the only covariates that correlated with disease-specific survival (see table 2).
Table 3 depicts the non-ocular primary cancers that patients with uveal melanoma developed during their lifetime, excluding metastases.
Compared with published SEER data for lifetime prevalences of these cancers,10 cutaneous melanoma was the only other malignancy with an elevated lifetime risk (p<0.0001).
The SEER database provides extensive epidemiological data regarding uveal melanoma. Histological subtype of uveal melanoma remains an important prognostic indicator. The SEER database indicates changes in the observed histopathological phenotypes with time. There was no clear pattern in the mixed epithelioid and spindle cell melanoma subtype over time, while the incidence of epithelioid tumours has remained steady. Interestingly, there appears to have been a substantial decrease in the proportion of spindle cell-type melanomas over the last four decades in the SEER cohort. This finding could be related to the trend of fewer enucleations performed for smaller, less aggressive tumours, with resultant skewing of the histological subtypes over time via sampling bias. However, the SEER database also allows for diagnosis by autopsy or biopsy, which would not be affected by treatment patterns. Another notable trend has been the increasing age at diagnosis of patients with uveal melanoma since 1973. This finding may be due to increasing life expectancy, changes in ophthalmic examination frequency in elderly patients (due to conditions such as macular degeneration and diabetes) or other occult causes. Other characteristics, such as race, gender and year of diagnosis, did not influence survival.
When examining the lifetime risk of developing another primary cancer, our analyses of the SEER data indicate that the lifetime risk of cutaneous melanoma is increased in patients with uveal melanoma, consistent with previous studies.11 ,12 However, the lifetime prevalence of other major tumour types does not appear to be significantly elevated. The lifetime prevalence of other primary cancers may be underestimated due to the lack of complete lifetime follow-up of all patients with uveal melanoma in the SEER database, especially those cases diagnosed in the latter portion of the study interval. However, the SEER data substantiate previous studies in that prostate and breast were also the most common post-treatment second cancer sites in the COMS trial.13 The COMS post-treatment non-ocular cancer incidence is most helpful for determining appropriate screening guidelines, but is subject to detection bias due to amplified screening after tumour diagnosis in a clinical trial. The lifetime rates of malignancies in patients with melanoma reported herein corroborate the possibility that patients with uveal melanoma may have an environmental or genetic predisposition to melanoma but perhaps not to other malignancies. These epidemiological findings, combined with histological and cytological assessments of biopsy samples,14 ,15 may assist in treatment planning and determining surveillance intervals in patients with uveal melanoma.
The SEER database provides extensive, well-validated cancer data. However, there are limitations to the available data; the SEER database does not include detailed descriptions of radiotherapy treatment protocols. Furthermore, the SEER programme does not include clinical examination data such as visual acuity. Lastly, the SEER programme lacks data from several large ocular oncology treatment centres in the USA, including those in Boston and Philadelphia. As selected geographic regions in the USA are currently captured by the SEER programme, this data set represents a very large, though not fully inclusive, cohort.
The present study examines epidemiological trends in uveal melanoma. The SEER data suggest several intriguing patterns in the evolution of this disease, including a difference in age at diagnosis between spindle and epithelioid tumours, an increasing age at diagnosis in patients with uveal melanoma and an elevated risk of cutaneous melanoma but no other cancer types in patients with uveal melanoma. These findings have implications for disease screening, for ocular as well as non-ocular malignancies, and potentially for therapeutic intervention and prognostic counselling.
This study used the SEER database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Surveillance, Epidemiology, and End Results (SEER) Programme Tumour Registry.
Contributors Design and conduct of the study: MTA and YIL. Collection, management, analysis and interpretation of the data: MTA, WFM and YIL. Preparation, review and approval of the manuscript: MTA, WFM and YIL.
Funding This work was supported in part by National Institutes of Health (NIH) Grant 1K12EY021475-01 and an unrestricted departmental grant from the Research to Prevent Blindness, New York, New York 10022, USA.
Competing interests None declared.
Ethics approval National Cancer Institute.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The National Cancer Institute SEER programme offers data to all researchers.