Uveal Melanoma: A Study on Incidence of Additional Cancers in the Swedish Population

Abstract

purpose. To investigate the occurrence of other primary malignancies before and after diagnosis of uveal melanoma in a Swedish population.

methods. In the Swedish Cancer Registry 2995 patients with uveal melanoma were notified during the period 1960 to 1998. In the same registry, a search for additional malignancies among these patients was performed. A matched case–control study with 2,916 patients and 14,577 population control subjects was set up for malignancies before diagnosis of uveal melanoma. Malignancies after diagnosis of uveal melanoma were evaluated in 2,995 patients through standardized incidence ratios (SIRs), based on the expected rates in the Swedish population.

results. Before the diagnosis of uveal melanoma, the odds ratio (OR) for the risk of cancer was 1.25 (95% CI: 0.98–1.59). No significantly increased risk was found for any specific malignancy. The OR for cutaneous melanoma was 1.74 (95% CI: 0.78–3.89). The risk of subsequent cancers was increased, SIR 1.13 (95% CI: 1.02–1.26). After reevaluation of archival specimens, the SIR of a cutaneous melanoma’s developing after a uveal melanoma was found to be 1.75 (95% CI: 0.87–3.12).

conclusions. An increased risk of second primary cancers was observed among Swedish patients with uveal melanoma. Metastases from uveal melanoma were found to be misclassified as cutaneous melanoma or as primary liver cancer.

The etiology and pathogenesis of uveal melanoma remains largely unknown. The role of ultraviolet (UV) radiation as a risk factor, or promoter for the development of uveal melanoma has been debated for decades, with both negative ^{1 2 3 4} and supportive reports, ^{5 6 7 8 9} as recently discussed in a review article by Singh et al. ¹⁰ In contrast to the increasing incidence rates of UV-inducible cutaneous melanoma apparent in white populations during the past decades, ^{11 12 13} the incidence of uveal melanoma has not increased. ^{14 15 16} The occurrence of additional primary tumors in patients with uveal melanoma or familiar clustering of cancers ^{17 18 19 20} may give indications of shared risk factors or a genetic predisposition, which could be further evaluated through case–control studies and, in recent years, through genetic linkage studies. 

In this study, we examined the association between the occurrence of uveal melanoma and additional cancers, by using reports to the national Swedish Cancer Registry filed during the period 1960 through 1998. 

Materials and Methods

The Swedish Cancer Registry, founded in 1958 is estimated to include more than 95% of the cancer cases in the population. ²¹ Physicians and pathologists give compulsory independent reports of newly detected tumors to the registry, which ensures a high inclusion rate. The 10-digit national registration number identifies each patient and avoids the risk of duplicate registration. For the whole period 1958 and onward, the site of tumor is available as the WHO ICD-7 code and the histopathology classification as the old histology code (WHO/HS/CANC/24.1). The registry does not accept death certificates only. ²² Included in the Swedish Cancer Registry are invasive solid malignancies, lymphomas, leukemia, and hematologic disorders such as reticulosis, myelofibrosis, and polycythemia vera. In situ malignant tumors of the breast, uterine cervix, gastrointestinal tract, skin, and prostate and some benign intracranial tumors are also included, but until recently (2003) basal cell carcinoma was not included. 

An investigation of the incidence of uveal melanoma in Sweden, based on reports to the Swedish Cancer Registry during the period January 1, 1960, to December 31, 1998, revealed 2997 cases in 2995 patients. ¹⁴ In the patients with uveal melanoma, up to five separate diagnoses of cancer were traced through the files in the Cancer Registry, along with the histology code and date of registration. Both the cancers occurring before the diagnosis of uveal melanoma and subsequent cancers were considered when investigating possible associations with uveal melanoma. The research protocol for the present study was approved by the Human Ethics Committee at the Karolinska Institute, in accordance with the statutes of the World Health Association’s Declaration of Helsinki. 

Primary Cancers before Uveal Melanoma

Previous cancers in any patient group is skewed in favor of malignancies with better prognosis compared to the cancers in the general population, which also contains individuals with lethal tumors, unable to develop ensuing cancers. Therefore, cancer cases notified before the diagnosis of uveal melanoma cannot be entered into an analysis based on expected rates from the general population. To evaluate the expected cancer incidence, we conducted a registry-based matched case–control study. Controls were drawn from the Registry of the Total Population, ²³ based on the annual Swedish census. For each case, five gender- and age (±1 day)-matched control subjects were selected, who were alive on the day the index person received a diagnosis of uveal melanoma. The study participants were the patients with uveal melanoma (n = 2,916, we could find no matching control subjects for 79 patients) and 14,577 control subjects. Through this procedure, the case patients and control subjects experienced the same time at risk. The Swedish Cancer Registry was searched for malignancies registered during the period from January 1, 1958 (or date of birth if born later) until the date of diagnosis of uveal melanoma for each index case and its corresponding control. Cases registered as cancer in situ were excluded. Conditional logistic regression analysis was used to calculate odds ratios (OR), as an estimation of relative risks, with a 95% confidence interval (CI). 

Cancers Diagnosed after Uveal Melanoma

The subsequent primaries after uveal melanoma were investigated in 2995 patients. The time at risk was calculated from the date of diagnosis of uveal melanoma until death or emigration occurred—otherwise until December 31, 1998. The expected number of malignancies, calculated as person-years at risk was estimated according to the incidence rates in the general Swedish population with respect to gender, age, and calendar period, using the software PYRS. ²⁴ The population was divided into four age groups (0–44, 45–59, 60–74, and 75+ years), four calendar periods (1960–1969, 1970–1979, 1980–1989, and 1990–1998), and by gender. The cases notified as in situ cancers were omitted in the analysis as the expected rates in the general population are based on invasive cancer. Cancer sites with five or more expected cases (in total) were further analyzed with standardized incidence ratios (SIRs). The SIR was calculated by dividing the observed number of cancer cases by the number of expected cases. The 95% CIs were estimated, assuming a Poisson distribution of the observed cases. 

In the case of registered primary cutaneous melanoma a further investigation proved necessary, as the tumor in several instances was reported in close conjunction with the death of the patient, suggesting a possible confounding. Available histologic specimens were re-evaluated by an experienced pathologist (SS). When the archival specimen could not be traced (mostly in cases diagnosed early during the investigation period) or if the patient survived only 2 years or less after the diagnosis of the second melanoma, the possibility of a metastasis from the uveal melanoma could not be ruled out, especially if the underlying cause of death was recorded as melanoma. Therefore, cases classified as having a primary uveal melanoma followed by a primary cutaneous melanoma either had a confirmative archival cutaneous melanoma specimen or survived more than 2 years after the diagnosis of cutaneous melanoma, suggesting that uveal melanoma metastasis to the skin was unlikely. As the incidence of primary liver cancer was remarkably high, available archival specimens were re-evaluated to exclude any misclassified metastases to the liver. This included, when possible, staining using a panel of the melanocytic markers HMB-45, Melan-A, and S-100. 

Results

Primary Cancers before Uveal Melanoma

Before the diagnosis of uveal melanoma, 154 patients had a registered cancer and, in total, 161 malignancies were found. In the control group, 631 patients experienced 661 cancers (Table 1) . This provided a tendency of an overall elevated risk of previous cancers in patients with uveal melanoma with an OR of 1.25 (95% CI: 0.98–1.59), compared with the population controls (Table 2) . No statistically significant increased risk was found in any specific group of cancer. Patients registered with cutaneous melanoma had an OR of 1.74 (95% CI: 0.78–3.89). 

Because 8 cutaneous melanomas were found among the cases and 23 in the control group, applying a post hoc analysis revealed that, to detect this difference (with a power of 80% and a two-sided α of 0.05), a sample size of 15,462 patients and 77,308 control subjects would be needed. 

The proportion of morphologically classified uveal melanomas (through local resection, primary or secondary enucleation) was 90.2% (n = 2704) in the Swedish uveal melanoma cohort, and a previous investigation of a random sample of these specimens (n = 916) revealed a misclassification rate of 0.33%, ¹⁴ which indicates that there was a small likelihood that a choroidal metastasis had been misclassified as a uveal melanoma. 

Cancers Diagnosed after Uveal Melanoma

During the period 1960 through 1998, the patients with uveal melanoma in the Swedish Cancer Registry accumulated 24,847 person-years of observation. Of the 2995 patients with uveal melanoma, 334 (11%) had one or more additional registrations of a primary cancer after the diagnosis of uveal melanoma. The in situ cases were not included in the analysis, as the expected rates are based on invasive cancers. Two additional registrations of a second cancer were found in 38 patients, and 1 patient had three cancers apart from the primary uveal melanoma. If multiple cancer registrations from the same site were found in a patient, only the first registration was put into analysis. The cancer sites were collapsed into the major ICD-7 groups. The observed cases as they appeared in the Swedish Cancer Registry and the expected number of malignant tumors are presented in Table 3 . After re-evaluation of archival specimens and adjustment for the probably misdiagnosed cases of cutaneous melanoma and primary liver cancer, the SIRs were calculated (Table 4) . 

An elevated overall risk was found for patients with uveal melanoma experiencing development of a second primary cancer, with an SIR of 1.13 (95% CI; 1.02–1.26). 

The evaluation of available archival specimens (n = 19) from the 29 patients registered with cutaneous melanoma after diagnosis of uveal melanoma revealed 9 with primary cutaneous melanoma. Ten of the specimens (registered as cutaneous melanoma) disclosed a melanoma metastasis to the skin, liver, or muscle tissue. Also, 18 patients died within 2 years after the diagnosis of the second melanoma, of which 7 had biopsy-validated liver metastases. With a high degree of certainty, 11 patients were considered to have primary uveal melanoma followed by primary cutaneous melanoma, with 9 patients with histopathology confirming primary cutaneous melanoma and 2 patients surviving 8 and 10 years, respectively, after diagnosis of cutaneous melanoma (because most patients with overt, disseminated uveal melanoma would have died after <2 years). When only confirmed cases were included, the risk of cutaneous melanoma developing after primary uveal melanoma was not found to be significantly elevated, (SIR 1.75; 95% CI: 0.87–3.12). Applying a post hoc analysis of the SIRs of primary cutaneous melanoma after uveal melanoma, our study of 2995 patients had a 53% power (two-sided α of 0.05) to detect the difference. To reach a power of 80% with a significance level of 0.05 would require 14 cases of primary cutaneous melanoma. 

A reduced risk of development of invasive nonmelanoma skin cancer was observed (SIR 0.68). In total, 10 cases were observed (14.7 expected), but an additional 14 in situ skin cancers were registered. 

Of 19 cases of registered primary liver cancer, 10 specimens were available for reassessment. Six specimens had paraffin blocks allowing for new sections to be stained with immunohistochemistry; 4 of these were found to be metastatic melanomas and 2 were confirmed to be primary hepatic cancers. An additional four specimens were not preserved in paraffin blocks. Reassessment on morphologic grounds confirmed two cases of bile duct cancer, but two cases featured undifferentiated cancer, in which metastatic melanoma could not be ruled out. When these percentages were extrapolated to the whole group of patients with registered primary liver cancers, 40% (n = 8) of the patients had a confirmed hepatic or bile duct cancer, 40% had metastatic melanoma, and 20% (n = 4) had undifferentiated cancers nonpermissive to classification. This suggests a corrected SIR of 0.86 (95% CI: 0.37–1.69) of primary liver cancer among the patients with uveal melanoma. 

A reduced risk of a primary lung cancer, with an SIR of 0.82 (95% CI: 0.50–1.28) was also observed. Both in the case of breast and ovarian cancers, the number of observed cases was below the expected numbers. 

Discussion

The use of nationwide, population-based registries when investigating possible associations between cancers is advantageous in reducing the biases of diagnosis and selection and provides better statistical power when a rare cancer, such as uveal melanoma, is studied. Previous studies using population-based databases, such as the Connecticut Tumor Registry, have pointed out increased risks (relative risk [RR] = 1.31) for development of a second primary cancer in patients who have cancer, where environmental exposures such as smoking and diet, treatment with chemotherapy or radiation, and a genetic predisposition for cancer may interact. ²⁵ In the same registry, eye cancer was associated with both an increased risk of a second cancer and cutaneous melanomas, but the results are not readily transferable to the subset of patients with uveal melanoma, as retinoblastoma and other eye malignancies were included in the analysis. ²⁶ In Finland, Teppo et al. ²⁷ found an increased risk for a second malignancy in patients who had the first cancer diagnosed before the age of 50, but not in the general population of persons with cancer. A study based on the Danish Cancer Registry ²⁸ found an elevated risk of second cancers in male patients with uveal melanoma, but not in females and no significantly raised risks for either cutaneous melanoma or nonmelanoma skin cancers. Case–control studies conducted in patients with uveal melanoma have produced inconsistent results. Turner et al. ²⁹ found an increased prevalence of non–basal-cell carcinomas and gynecological cancers, whereas Lischko et al. ³⁰ and Holly et al. ³¹ found nonsignificant increased rates of both skin- and non-skin cancers in patients with uveal melanoma, compared with those in control subjects. Concurrent second cancers in patients with uveal melanoma have been reported to occur in 8% to 13%. ^{30 31 32 33}  

Our findings, with data from a nationwide, population-based survey of patients with uveal melanoma over a 39-year period and with minimal loss to follow-up (due to emigration in nine cases), indicate that uveal melanoma is associated with a raised risk of a second primary cancer. According to the risk ratios, 13% more cancer cases occurred among Swedish patients with uveal melanoma compared with the expected rates in the general population. 

The risk of actual development of two primary cancers is most variable, depending on factors such as applied treatment, curability, and age at diagnosis of the first cancer. The median age at diagnosis of uveal melanoma in our survey was 64 years for both men and women, and most of the patients underwent primary enucleation (86.4%). Brachytherapy (ruthenium episcleral plaque) was applied in 12.9% of the cases and 0.7% received therapy with charged particles (protons). The use of chemotherapy in patients with uveal melanoma has been restricted to palliation, and because of the short life expectancy of patients with metastases the risk of inducing another cancer is probably negligible. In the setting of uveal melanoma, it is therefore unlikely that the management would inflict an elevated risk of second primary cancer. 

Our initial findings of a substantially raised risk of primary liver cancer are concordant with the results (SIR 5.10) Swerdlow et al. ²⁸ reported from the Danish Cancer Registry. Although, after reevaluation of the archival specimens the SIR was adjusted to 0.86 (95% CI: 0.37–1.69) as primary liver cancer was only confirmed in 4 of 10 specimens whereas 4 of 10 specimens were melanoma metastases. 

An association between breast cancer and uveal melanoma has been suggested, the proposed linkage through BRCA2 gene mutations. ^{34 35} In patients with uveal melanoma, the prevalence of BRCA2 mutations are rare, probably not more than 2% to 3%. ^{36 37} In the Swedish patients with uveal melanoma, we found a relative risk of female breast cancer of 0.92 (95% CI: 0.61–1.33). As expected, the identification of patients possibly carrying a rare mutation would be diluted in a population-based investigation. 

Although both cutaneous and uveal melanocytes are derived from the neural crest, the malignant melanomas arising in the skin and uvea display discrepancies, both in their clinical behavior and on a genetic and molecular level. ^{38 39 40 41 42 43} Solar light is established as one causative factor in cutaneous melanoma ⁴⁴ but the role of UV light as a risk factor for uveal melanoma is questionable. ¹⁰ An association has been proposed between the two entities through the dysplastic nevus syndrome (DNS), also called familial atypical mole-and-melanoma (FAM-M) syndrome, an autosomal dominant condition predisposing the carrier to cutaneous melanoma. The definition of DNS has varied in the literature and consequently the prevalence, which makes it difficult to compare the investigations in which dysplastic nevi and cutaneous melanoma were reported to occur more often in patients with uveal melanoma and their close relatives than by chance alone. ^{45 46 47 48 49} In a previous case–control study, we did not detect a higher frequency of uveal nevi in patients with DNS. ⁵⁰ Case reports of coexistent primary uveal and cutaneous melanoma have been scarce, ^{48 51 52} but a recent population-based investigation by Shors et al. ⁵³ using the SEER database pointed out a 4.6-fold increased risk of development of cutaneous melanoma in the presence of an initial uveal melanoma, findings that were supported by information from the Swedish Family-Cancer database, with a significant SIR of 5.04 for development of cutaneous melanoma 1 to 10 years after an uveal melanoma. ⁵⁴ The reverse, an increased risk of uveal melanoma after a primary cutaneous melanoma was not found in either study. 

In our investigation, to minimize the risk of including uveal melanoma metastases, we omitted the registered cases of cutaneous melanoma (18 patients) diagnosed only at autopsy or within 2 years of the patient’s death, which still leaves an SIR of 1.75. If we had calculated the standardized incidence ratios directly from the notifications in the Cancer Registry, and disregarded the possibility of a miscoded metastatic uveal melanoma, the SIR would be 4.60 (95% CI: 3.08–6.61). This finding could indicate that previously published SIRs in patients with uveal melanoma may have included cases in which metastatic uveal melanoma was miscoded as primary cutaneous melanoma. 

Although misclassifications of metastatic uveal melanomas in cancer registries probably have inflated incidence ratios, coexistence of uveal and cutaneous melanoma appears to be more common than previously believed. Using stringent criteria, we could not confirm that patients with uveal melanoma are at a statistically significant increased risk of development of subsequent cutaneous melanoma. However, lesions reported as subsequent primary cutaneous melanoma but unavailable for histopathological confirmation (and therefore withdrawn from analysis in our study) may have included true cases of primary cutaneous melanoma. Our findings could provide some support for dermatological examination after diagnosis of uveal melanoma. 

 Supported by grants from Crown Princess Margaretha’s Foundation for the Visually Impaired, Stockholm, Sweden; Synfrämjandets Research Fund, Stockholm, Sweden; and the St. Erik Foundation for Research in Ophthalmology, Stockholm, Sweden.

 Submitted for publication July 11, 2005; revised September 14, 2005; accepted November 28, 2005.

 Disclosure: L. Bergman, None; B. Nilsson, None; B. Ragnarsson-Olding, None; S. Seregard, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Cancers Found Prior to Diagnosis of Uveal Melanoma

ICD7	Site	Cases	Control Subject
140	Lip	4	6
142	Salivary glands	2	6
143–144	Mouth	—	6
145–148	Pharynx	1	3
150	Esophagus	—	2
151	Stomach	2	19
152	Small intestíne	2	6
153	Colon	12	59
154	Rectum and anus	5	28
155	Primary liver, bile duct	—	4
157	Pancreas	—	5
160	Nose and sinuses	1	—
161	Larynx	—	9
162–163	Lung	2	25
170	Breast	25	123
171	Cervix	8	24
172–174	Uterus	7	25
175	Ovary	1	24
176	Female genital organs	3*	4
177	Prostate	24	80
178	Testis	1	2
179	Male genital organs	2	1
180	Kidney	8	15
181	Other urinary organs	8	37
190	Cutaneous melanoma	8	23
191	Cutaneous (nonmelanoma)	9	28
193	Nervous system	6	22
194	Thyroid gland	4	7
195	Other endocrine glands	7	20
196	Bone	—	2
199	Unspecified sites	—	1
200	Non-Hodgkin lymphoma	4	22
201	Hodgkin disease	2	4
202	Reticulosis	1	1
203	Myeloma	—	5
204–207	Leukemias	4	5
208	Polycythemia vera	—	5
209	Myelofibrosis	1	—

  n = 2,916 patients with uveal melanoma and in 14,577 age- and sex-matched control subjects from the general Swedish population

*  Including a malignant melanoma of the vulva.

Table 2.

 

OR with 95% CI for Previous Cancers in Various Organ Systems

ICD-7	Site of Previous Cancer	Cases	Control Subjects	OR	95% CI	P
140–209	All sites	161	661	1.25	0.98–1.59	0.07
140–150	Lip-esophagus	7	23	1.52	0.65–3.54	0.33
153–154	Colon-anus	17	87	0.97	0.57–1.63	0.89
170	Breast	25	123	1.02	0.66–1.58	0.94
171–176	Female genital organs	19	77	1.24	0.75–2.05	0.41
177	Prostate	24	80	1.52	0.96–2.43	0.08
180–181	Urinary system	16	52	1.32	0.66–2.64	0.44
190	Cutaneous melanoma	8	23	1.74	0.78–3.89	0.18
191	Cutaneous (nonmelanoma)	9	28	1.62	0.76–3.35	0.23
194–195	Endocrine glands	11	27	1.88	0.50–7.07	0.35
200–209	Lymphoma/leukemia	12	42	1.46	0.69–3.08	0.32

  n = 2,916 patients with uveal melanoma; 14,577 age- and sex-matched control subjects.

Table 3.

 

Registrations of Subsequent Primary Malignancies

ICD7	Site	Total Expected	Total Observed
	All sites	306.1	374
140	Lip	1.7	2
141	Tongue	0.4	0
142	Salivary glands	0.2	1
143–144	Mouth	0.7	0
145–148	Pharynx	0.1	4
150	Esophagus	2.8	2
151	Stomach	16.6	14
152	Small intestine	1.3	2
153	Colon	25.5	30
154	Rectum and anus	14.5	16
155	Primary liver, bile duct	9.3	19
157	Pancreas	10.3	14
160	Nose and sinuses	0.1	2
161	Larynx	1.3	1
162–163	Lung	23.1	19
170	Breast	30.4	28
171	Cervix	2.5	4
172–174	Uterus	7.1	10
175	Ovary	6.3	6
176	Female genital organs	1.1	2
177	Prostate	49	54
178	Testis	0	0
179	Male genital organs	0.1	1
180	Kidney	9.7	10
181	Other urinary organs	16	19
190	Cutaneous melanoma	6.3	29
191	Cutaneous (nonmelanoma)	14.7	10
193	Nervous system	6.7	10
194	Thyroid gland	1.4	3
195	Other endocrine glands	3.7	6
196	Bone	0	1
199	Unspecified sites	10.3	19
200	Non-Hodgkin lymphoma	7.9	14
201	Hodgkin disease	0.6	0
203	Myeloma	4.4	9
204–207	Leukemias	6.1	8
208	Polycythemia vera	0.8	2
209	Myelofibrosis	0.2	3

  n = 2,995 patients with uveal melanoma patients as notified in the Swedish Cancer Registry during the period 1960–1998.

Table 4.

 

SIR with 95% CI for Subsequent Cancers

ICD7	Site	Total Expected	Total Observed	SIR	95% CI
	All sites*	306.1	346	1.13	1.02–1.26
151	Stomach	16.6	14	0.84	0.46–1.42
153	Colon	25.5	30	1.18	0.79–1.68
154	Rectum and anus	14.5	16	1.10	0.63–1.79
155	Primary liver, bile duct*	9.3	8	0.86	0.37–1.69
157	Pancreas	10.3	14	1.36	0.74–2.28
162–163	Lung	23.1	19	0.82	0.50–1.28
170	Breast	30.4	28	0.92	0.61–1.33
172–174	Uterus	7.1	10	1.41	0.68–2.59
175	Ovary	6.3	6	0.95	0.35–2.07
177	Prostate	49	54	1.10	0.83–1.44
180	Kidney	9.7	10	1.03	0.49–1.90
181	Other urinary organs	16	19	1.19	0.57–1.62
190	Cutaneous melanoma*	6.3	11	1.75	0.87–3.12
191	Cutaneous (nonmelanoma)	14.7	10	0.68	0.33–1.25
193	Nervous system	6.7	10	1.49	0.72–2.74
194–195	Thyroid+endocrine glands	5.1	9	1.76	0.81–3.35
199	Unspecified sites	10.3	19	1.84	1.11–2.88
200	Non-Hodgkin lymphoma	7.9	14	1.77	0.97–2.97
204–207	Leukemias	6.1	8	1.31	0.57–2.58

  n = 2995 Swedish patients with uveal melanoma.

*  Adjusted after re-evaluation of archival specimens.