Study Population

This retrospective cohort study followed a group of patients with a diagnosis of uveal melanoma as their first malignancy, who were identified in the Surveillance, Epidemiology, and End Results (SEER) 9 database from 1973 to 2011. The SEER program collects data on all persons diagnosed with cancer residing in several geographically defined regions—the SEER 9 database includes 9 registries, covering approximately 10% of the United States (US) population and being comparable to the US general population, with a case ascertainment rate of 98%. Despite including fewer registries than the newer SEER databases, SEER 9 is the only one that provides data from 1973 onwards. This study met the definition of exempt research. As a result, Institutional Review Board approval was not required. The SEER data is considered nonhuman subject research. The data is anonymized. No patient contact is involved in its collection. The collection of SEER data is also exempt from Health Insurance Portability and Accountability Act regulations.

Patient selection was based on the primary site and histologic subtype of the first malignancy, according to the International Classification of Diseases for Oncology, third edition (ICD-O-3), as follows: topography codes C69.2 (retina), C69.3 (ciliary body), and C69.4 (choroid) for uvea; morphology codes 8720.3 to 8774.3 for melanoma. As previously described, most “retinal” melanomas probably represent misclassification of uveal melanoma; thus they were also considered. ICD-O-3 includes the iris in ciliary body topography. Therefore, these cases were not analyzed separately. SPNs were considered as those diagnosed at least 2 months after the uveal melanoma, as defined by the SEER program. We excluded cases where the diagnosis of uveal melanoma was only made by death certificate or autopsy, because they cannot contribute with person-time for SPN assessment. For both primary and SPNs, we selected only cases with malignant behavior. Regarding treatment for the first malignancy, the SEER database includes information on the type of radiotherapy provided (ie, beam radiation, radioactive implants/radioisotopes, combination of both, radiation not specified or unknown). No details are available for the type of beam irradiation applied and for the doses of any of these treatment modalities. Additionally, there are also no data in the SEER database on the type of surgery performed (ie, enucleation or other surgical modalities) for eye malignancies.

Statistical Analysis

We used the SEER*Stat program – MP-SIR session (Surveillance Research Program, National Cancer Institute SEER*Stat software; seer.cancer.gov/seerstat , version 8.1.5) to perform the multiple primary analyses. Person-years at risk were determined starting in the third month after uveal melanoma diagnosis until date of death, last follow-up, SPN diagnosis, or end of 2011, whichever came first. Observed number of cancers corresponded to the actual reported SPN cases. The expected number of SPNs corresponded to those that would develop if the study cohort patients experienced the same rate of occurrence as a reference SEER population, with comparable sex, 5-year age group, race (white/unknown, black, other), site, and calendar year. For this analysis, we used the reference population “Incidence – SEER 9 Regs Research Data, Nov 2013 Sub (1973–2011) <Katrina/Rita Population Adjustment> – Linked To County Attributes – Total U.S., 1969–2012 Counties.” Briefly, this reference population includes all people living in the SEER 9 geographic areas between 1973 and 2011, based on the US Census Bureau’s Population Estimates Program External Web Site Policy, in collaboration with the National Center for Health Statistics, and with support from the NCI through an interagency agreement. Referent rates (for expected cases) represent SEER 9 incidence rates, which are then stratified by age, sex, race, calendar year, and cancer site, according to the characteristics of our cohort of uveal melanoma patients.

Standardized incidence ratios (SIR) were obtained by observed-to-expected ratio and the 95% confidence interval (95% CI) was calculated based on the Poisson distribution. The excess absolute risk of SPN was calculated by subtracting the expected number from the number of observed cases and dividing by the person-years at risk, and corresponded to the additional cancers (beyond the expected amount). It was expressed per 10 000 person-years.

Univariate and multivariate Cox proportional hazards regression models were used to evaluate the effect of radiotherapy on SPN risk. Owing to the small number of nonwhite subjects in our cohort, for these analyses race was coded as a dichotomous variable (white/unknown vs black/other). A secondary analysis investigated the effect of the type of radiotherapy (beam radiation, radioactive implants/radioisotopes) on SPN risk. For this analysis, subjects given a combination of radiotherapy or with unknown type of radiotherapy were excluded. For multivariate analysis, the proportional hazards assumption was tested using the log minus log curve {ln[ _ ln ( S )]} and the goodness-of-fit test, and no violation was observed for any of the variables. Stata version 12.1 (StataCorp LP, College Station, Texas, USA) was used for all these Cox analyses. The previous analyses censored all subjects who died at the time of death, but an alternative approach could consider death from uveal melanoma or death from other cause as a competing risk. Therefore, all of the previous multivariate models were refit accounting for competing risks using the stcompet and stcrreg routines in Stata.

Study Population

The study cohort consisted of 3976 patients diagnosed with uveal melanoma as their first malignancy ( Table 1 ). Most cases were coded as malignant melanoma without other histologic specification (n = 2475, 62%). The remaining cases were classified as spindle cell melanoma (n = 768, 19%), epithelioid cell melanoma (n = 192, 5%), mixed epithelioid and spindle cell melanoma (n = 504, 13%), amelanotic melanoma (n = 32, 0.8%), and others (n = 5, 0.1%). The median follow-up time was 83 months, with 62% (n = 2427) of the subjects followed for at least 5 years (range, 1–463 months). Nearly 30% (n = 1168) of the patients died of uveal melanoma and the 5- and 10-year disease-specific survival rates were 80% and 69%, respectively. The global survival rate during the available follow-up was 44% (n = 1740).

Secondary Cancers After Uveal Melanoma

SPN occurred in 641 of the patients with uveal melanoma (16%), within a median time of 75 months (range, 2–425 months). Most (n = 583, 15%) had 1 additional cancer and a few had 2 (n = 57, 1%) or 3 (n = 1, 0.03%) new primaries. SIRs demonstrated that patients with uveal melanoma have a significant 11% excess risk of SPN when compared to the reference population, with an excess absolute risk of 17.28/10 000 person-years. This was mainly owing to a significantly higher risk of tumors of the salivary glands, kidney, and soft tissues, as well as skin melanomas ( Table 2 ). The median age at the time of diagnosis of these SPNs was 77, 67, 81, and 69 years, respectively.

A statistically significant increased risk for a second uveal melanoma was also noticed in our study cohort, with 13 cases reported ( Table 3 ) within a median latency time of 7 years (range, 2 months to 27 years). Two of these cases occurred in the fellow eye, 2 months and 7 years later than the first malignancy. Ipsilateral second melanomas were coded as new primaries in 9 patients and in 2 patients data regarding laterality were not available. Most cases of second uveal melanomas (n = 9, 69%) occurred after an initial treatment with radiotherapy. As discussed below, we considered that at least some of these second tumors likely represent misclassified recurrences. Therefore, we recalculated the primary outcomes with these second malignancies excluded. Our results confirmed that patients with uveal melanoma demonstrated an increased risk of SPN for all sites of 1.09 (95% CI: 1.01–1.18), excluding cases coded as a second uveal melanoma ( Table 2 ).

When evaluating the risk during the follow-up period ( Table 4 ), we observed that the increased risk for salivary gland tumors was only significant within the first 5 years after the diagnosis of uveal melanoma. Kidney tumors occurred more frequently in 2 periods: in the first 5 years of follow-up and later among 10- to 19-year survivors. Skin melanoma risk was increased throughout, and was still significant in 19-year survivors.

Considering the relevance of age and sex for cancer in general, we additionally evaluated SIRs stratified by these parameters. Our results revealed that subjects aged 30–59 years at the time of diagnosis of uveal melanoma presented the higher risk of developing SPN (for all sites, SIR = 1.18; 95% CI: 1.04–1.34). Interestingly, those diagnosed before that age did not present a significantly increased risk compared to the reference population ( Table 5 ). Regarding sex, compared to the female reference population, women with a previous uveal melanoma presented a 17% significantly increased risk of all sites SPN (95% CI: 1.04–1.32, Table 6 ). There was a trend towards an increased risk for men (7%), but it was not significant. For both sexes, the described significantly higher risk of skin melanoma and kidney tumors was still observed.

Stay updated, free articles. Join our Telegram channel

Join