Volume 103, Issue 5 p. 1000-1007
Original Article
Free Access

Incidence of noncutaneous melanomas in the U.S.

Colleen C. McLaughlin M.P.H.

Corresponding Author

Colleen C. McLaughlin M.P.H.

New York State Cancer Registry, New York State Department of Health, Albany, New York

Fax: (518) 473-6789

New York State Cancer Registry, New York State Department of Health, Corning Tower Room 536, Empire State Plaza, Albany, NY 12237===Search for more papers by this author
Xiao-Cheng Wu M.P.H.

Xiao-Cheng Wu M.P.H.

Louisiana Tumor Registry and Epidemiology Program, School of Public Health, Louisiana State University Health Sciences Center, New Orleans, Louisiana

Search for more papers by this author
Ahmedin Jemal D.V.M., Ph.D.

Ahmedin Jemal D.V.M., Ph.D.

Epidemiology and Surveillance Research, American Cancer Society, Atlanta, Georgia

Search for more papers by this author
Howard J. Martin Ph.D.

Howard J. Martin Ph.D.

Virginia Cancer Registry, Virginia Department of Health, Richmond, Virginia

Search for more papers by this author
Lisa M. Roche M.P.H., Ph.D.

Lisa M. Roche M.P.H., Ph.D.

Cancer Surveillance Program, Cancer Epidemiology Services, New Jersey Department of Health and Senior Services, Trenton, New Jersey

Search for more papers by this author
Vivien W. Chen Ph.D.

Vivien W. Chen Ph.D.

Louisiana Tumor Registry and Epidemiology Program, School of Public Health, Louisiana State University Health Sciences Center, New Orleans, Louisiana

Search for more papers by this author
First published: 13 January 2005
Citations: 504

Abstract

BACKGROUND

Description of the epidemiology of noncutaneous melanoma has been hampered by its rarity. The current report was the largest in-depth descriptive analysis of incidence of noncutaneous melanoma in the United States, using data from the North American Association of Central Cancer Registries.

METHODS

Pooled data from 27 states and one metropolitan area were used to examine the incidence of noncutaneous melanoma by anatomic subsite, gender, age, race, and geography (northern/southern and coastal/noncoastal) for cases diagnosed between 1996 and 2000. Percent distribution by stage of disease at diagnosis and histology were also examined.

RESULTS

Between 1996 and 2000, 6691 cases of noncutaneous melanoma (4885 ocular and 1806 mucosal) were diagnosed among 851 million person-years at risk. Ocular melanoma was more common among men compared with women (6.8 cases per million men compared with 5.3 cases per million women, age-adjusted to the 2000 U.S. population standard), whereas mucosal melanoma was more common among women (2.8 cases per million women compared with 1.5 cases per million men). Rates of ocular melanoma among whites were greater than eight times higher than among blacks. Rates of mucosal melanoma were approximately two times higher among whites compared with blacks.

CONCLUSIONS

In contrast to cutaneous melanoma, there was no apparent pattern of increased noncutaneous melanoma among residents of southern or coastal states, with the exception of melanoma of the ciliary body and iris. Despite their shared cellular origins, both ocular and mucosal melanomas differ from cutaneous melanoma in terms of incidence by gender, race, and geographic area. Cancer 2005. © 2005 American Cancer Society.

Relative to melanoma of the skin, noncutaneous melanomas are rare. Of 82,943 cases of melanoma with a known primary site reported to the National Cancer Data Base (NCDB) in the United States for 1985–1994, only 5.5% were ocular and 1.3% occurred at mucosal sites.1 Because of this rarity, there have been few descriptive studies of mucosal melanomas in the United States or internationally. Published estimates of the incidence of ocular melanoma in the United States from the National Cancer Institute's (NCI) Surveillance, Epidemiology, and End Results (SEER) program were based on approximately 100 cases per year, covering 10% of the U.S. population.2 A recent population-based description of the incidence of ocular melanoma in Australia was also based on approximately 750 cases, with approximately 125 cases reported per year.3 Use of the combined high-quality data from the North American Association of Central Cancer Registries (NAACCR), covering approximately 62% of the U.S. population, offers a unique opportunity to examine incidence patterns of noncutaneous melanoma by subsite, age, gender, race, histology, summary stage, and region within the United States. Although the current article focuses on noncutaneous melanoma, data for cutaneous melanomas are also presented to highlight the similarities and contrast the differences in these diseases, which may improve the understanding of both malignancies.

MATERIALS AND METHODS

Incidence data were obtained from the NAACCR analytic file, December 2002 data submission. These data cover patients with cancer diagnosed between 1996 and 2000. The criteria for inclusion in the NAACCR analytic file are 1) having a completeness estimate of ≥ 90% for each of the 5 years from 1996 to 2000; 2) having < 0.1% duplicate records for 1996–2000; 3) having < 2% missing data for gender, county of residence, and age and < 5% missing race; 4) having < 5% of cancer cases in each year reported from death certificates only; and 5) having data for 1996–2000 available within 23 months of the close of the 2000 diagnosis year (December 2002). Most registries included in the study met stricter criteria of 95% completeness of case ascertainment and < 3% of cases from death certificates only. Not all registries that met these criteria are included because member registries had to provide explicit consent for inclusion of their data in the current study. Twenty-seven states and one metropolitan area met these inclusion criteria and agreed to participate in the study: Alaska, Arizona, California, Colorado, Connecticut, Hawaii, Idaho, Illinois, Iowa, Kentucky, Louisiana, Michigan, Minnesota, Montana, Nebraska, New Jersey, New Mexico, North Carolina, Oregon, Pennsylvania, Rhode Island, South Carolina, Utah, Washington, West Virginia, Wisconsin, Wyoming, and Metropolitan Atlanta in Georgia.

Melanomas were identified based on International Classification of Diseases for Oncology, 2nd edition (ICD-O-2), morphology codes 8720 through 8790. Anatomic site was identified based on ICD-O-2 topography codes. Ocular melanomas (ICD-O-2 C69) were subdivided further into uveal melanomas, which included choroid (C69.3) and iris and ciliary body (C69.4), and other ocular melanomas, which are predominantly conjunctival (C69.0), but also include other parts of the eye (e.g., the retina) as well as overlapping and unspecified ocular subsites. For mucosal melanomas, anatomic site was subdivided further into anorectal (C20-C21), genital (C51-C63), nasal cavity and accessory sinuses (C30-C31), and oral cavity (C00-C14). In the ICD-O-2 classification scheme, the skin of the anus and perianal skin are considered to be cutaneous. Cutaneous melanoma, ICD-O-2 code C44, includes all skin except the skin of the vulva, penis, and scrotum, which are considered to be mucosal sites in the current study. Melanomas with unknown primary site (< 0.04% of melanomas) were also considered to be cutaneous. For the anatomic sites included in our study, the analytic file for NAACCR includes only tumors with invasive behavior.

Average annual age-adjusted incidence rates by gender, race, anatomic site, and region within the United States were calculated using the 2000 U.S. population standard. Rates standardized to the world standard also were calculated for all races and for persons classified as white to facilitate international comparisons. Confidence intervals for the age-adjusted rate ratios were calculated using the method from Miettinen.4 Population estimates for the cancer registries included in the analysis were obtained from the NCI.5 All analyses were conducted in SEER*Stat version 5.1.14.6 Because of the rarity of noncutaneous melanoma, all rates are expressed as per million population per year. In addition to rates, the percent distribution of anatomic site, histology, and summary stage was calculated. Summary stage was grouped into localized, regional spread, and distant metastases.7 Summary stage was not available for cases diagnosed in California.

To examine the possibility of differences in noncutaneous melanoma incidence by latitude or coastal location, the states included in the analysis were divided as follows. States were considered southern if the population-weighted centroid of the state was on or below the population weighted median line for the United States, based on data from the U.S. Census Bureau.8 The southern region included Arizona, Atlanta, GA, California, Hawaii, Kentucky, Louisiana, New Mexico, North Carolina, South Carolina, and West Virginia. The northern region included Alaska, Colorado, Connecticut, Idaho, Illinois, Iowa, Michigan, Minnesota, Montana, Nebraska, New Jersey, Oregon, Pennsylvania, Rhode Island, Utah, Washington, Wisconsin, and Wyoming. States were considered coastal if they bordered on the Atlantic Ocean, the Pacific Ocean, or the Gulf of Mexico, except for Alaska, which was considered noncoastal. The coastal region included California, Connecticut, Hawaii, Louisiana, New Jersey, North Carolina, Oregon, Rhode Island, South Carolina, and Washington. The noncoastal region included Alaska, Arizona, Atlanta, GA, Colorado, Idaho, Illinois, Iowa, Kentucky, Michigan, Minnesota, Montana, Nebraska, New Mexico, Pennsylvania, Utah, West Virginia, Wisconsin, and Wyoming. Alaska was considered noncoastal because of the limited amount of beach recreation in that state, although repeating the analysis with Alaska considered coastal had very little impact on the results. For these region-specific analyses, rates were calculated among whites only to avoid confounding by race.

RESULTS

From 1996 to 2000, 6691 noncutaneous melanomas were diagnosed over 851 million persons-years of observation, with an age-adjusted rate of 8.2 noncutaneous melanoma cases per million residents per year, adjusted to the 2000 U.S. population standard (Table 1). Ocular and mucosal melanomas accounted for 3.7% and 1.4% of the 133,209 melanoma cases reported during this time period, respectively. Rates age adjusted to the world standard for all races and for whites only are presented in Table 2.

Table 1. Age-Adjusted (2000 U.S. Standard) Incidence Ratesa of Melanoma by Gender and Subsite, United States, 1996–2000
Site Male and female Male Female
Total no. of cases Rate per million 95% CI Total no. of cases Rate per million 95% CI Total no. of cases Rate per million 95% CI
All melanoma 133,209 161.7 160.2–162.6 74,296 202.0 200.6–203.5 58,913 133.2 132.1–134.3
Melanoma of the skin 126,518 153.5 152.7–154.3 71,255 193.7 192.2–195.1 55,263 125.2 124.1–126.2
All noncutaneous sites 6691 8.2 8.0–8.4 3041 8.4 8.1–8.7 3650 8.1 7.8–8.3
Ocular 4885 6.0 5.8–6.1 2514 6.8 6.6–7.1 2371 5.3 5.1–5.5
 Uveal 4030 4.9 4.8–5.1 2088 5.7 5.4–5.9 1942 4.4 4.2–4.6
  Choroid 3477 4.3 4.1–4.4 1809 4.9 4.7–5.1 1668 3.7 3.6–3.9
  Iris and ciliary body 553 0.7 0.6–0.7 279 0.8 0.7–0.9 274 0.6 0.5–0.7
 Nonuveal 855 1.0 1.0–1.1 426 1.2 1.1–1.3 429 0.9 0.9–1.0
  Conjunctiva 324 0.4 0.4–0.4 161 0.4 0.4–0.5 163 0.4 0.3–0.4
Mucosal 1806 2.2 2.1–2.3 527 1.5 1.4–1.7 1279 2.8 2.6–2.9
 Nasal cavity 255 0.3 0.3–0.4 120 0.4 0.3–0.4 135 0.3 0.2–0.3
 Accessory sinuses 140 0.2 0.1–0.2 69 0.2 0.2–0.3 71 0.2 0.1–0.2
 Oral cavity 164 0.2 0.2–0.2 91 0.3 0.2–0.3 73 0.2 0.1–0.2
 Anorectal 299 0.4 0.3–0.4 104 0.3 0.2–0.4 195 0.4 0.4–0.5
 Genital tract 776 1.0 0.9–1.0 53 0.2 0.1–0.2 723 1.6 1.5–1.7
  • 95% CI: 95% confidence interval.
  • a Rates are age adjusted to the 2000 U.S. standard population, with estimated 95% CIs, Rates and 95% CI values are rounded.
Table 2. Age-Adjusted (World Standard) Incidence Ratesa of Noncutaneous Melanoma by Gender and Subsite, United States, 1996–2000
Site/race group Male and female Male Female
Total no. of cases Rate per million 95% CI Total no. of cases Rate per million 95% CI Total no. of cases Rate per million 95% CI
All noncutaneous sites
 All races 6691 5.6 5.5–5.8 3041 5.7 5.5–6.0 3650 5.5 5.4–5.8
 Whites only 6301 6.1 6.0–6.3 2884 6.2 6.0–6.5 3417 6.1 5.8–6.3
Ocular melanomas
 All races 4885 4.2 4.1–4.4 2514 4.8 4.6–5.0 2371 3.8 3.6–4.0
 Whites only 4676 4.7 4.6–4.9 2410 5.3 5.1–5.5 2266 4.2 4.0–4.4
Mucosal melanomas
 All races 1806 1.4 1.3–1.5 527 0.9 0.9–1.0 1279 1.8 1.7–1.9
 Whites only 1625 1.4 1.4–1.5 474 1.0 0.9–1.1 1151 1.8 1.7–2.0
  • 95% CI: 95% confidence interval.
  • a Rates are age adjusted to the world standard population, with estimated 95% CIs. Rates and 95% CI values are rounded.

The age-adjusted (U.S. standard) rates of noncutaneous melanoma among men (8.4 cases per million) and women (8.1 cases per million) were similar, although the distribution of anatomic site of melanoma differed by gender (Table 1). Among men, 82.7% of noncutaneous melanomas were ocular, whereas among women, 65.0% were ocular. The ocular melanoma rate among men (6.8 cases per million men per year) was 28.3% higher than among women (5.3 cases per million women per year). The male-to-female rate ratio was 1.29, with a 95% confidence interval (95% CI) of 1.22–1.36. Most ocular melanomas occurred in the uveal tract (82.5%), of which 86.3% occurred in the choroid (Table 1). Melanomas in other parts of the eye were very rare, with annual age-adjusted rates of < 2 cases per million residents per year among both men and women (2000 US standard).

Unlike ocular melanomas, the mucosal melanoma rate among women (2.8 cases per million women per year) was 86.7% higher than the corresponding rate among men (1.5 cases per million men per year). The male-to-female rate ratio was 0.54, with a 95% CI of 0.50–0.61. This higher rate among women was caused by higher rates of genital tract melanomas, which among women are predominantly melanomas of the vulva and vagina and account for 56.5% of the mucosal melanomas among women. Among women, vulvar melanomas accounted for 76.7% of genital melanomas (555 cases, age-adjusted rate 1.2 cases per million, 95% CI = 1.1–1.3) and vaginal melanomas accounted for 19.8% of genital melanomas (143 cases, age-adjusted rate 0.3, 95% CI = 0.3–0.4). Sixteen cases of melanoma of the cervix were reported. Among men, melanomas of the penis were the most common genital melanomas (35 cases, age-adjusted rate 0.1, 95% CI = 0.07–0.14). Seventeen cases of melanoma of the scrotum also were reported. For nongenital sites, no differences in rates of mucosal melanoma were observed by gender.

Melanomas of the head and neck were the most common mucosal sites among men and the second most common among women, with the nasal cavity being the most common head and neck site among both genders. Within the oral cavity, the most common sites among men and women combined were the palate (40 cases), nasopharynx (26 cases), lip (24 cases), gum (23 cases), and salivary gland (18 cases). Anorectal melanomas accounted for 16.5% of mucosal melanomas, and the rates were slightly higher among women than among men (male-to-female rate ratio 0.75, 95% CI = 0.57–0.90).

The age distribution of noncutaneous melanomas varied by gender and anatomic site (Table 3). Among women, 65.5% of noncutaneous melanomas occurred after the age of 60, compared with 58.7% among men, reflecting the older age distribution of women in the United States as well as the older age distribution of mucosal melanomas compared with ocular melanomas. The rate of ocular melanoma was higher among men compared with women in every age group. The rates of mucosal melanomas were higher among women compared with men in every age group, primarily because of the higher rates of genital melanomas.

Table 3. Age-Specific Incidence Ratesa of Noncutaneous Melanoma by Gender and Subsite, United States, 1996–2000
Age group (yrs) All noncutaneous melanomas Ocular Mucosal
Total no. of cases (%) Rate per million 95% CI Total no. of cases (%) Rate per million 95% CI Total no. of cases (%) Rate per million 95% CI
Males
 00–29 77 (2.5) 0.4 0.3–0.5 65 (2.6) 0.3 0.3–0.4 <16 (2.8)
 30–39 184 (6.1) 2.7 2.3–3.1 165 (6.6) 2.4 2.0–2.8 19 (3.6) 0.3 0.2–0.4
 40–49 396 (13.0) 6.3 5.7–7.0 338 (13.4) 5.4 4.8–6.0 58 (10.9) 0.9 0.7–1.2
 50–59 598 (19.7) 14.2 13.1–15.4 517 (20.6) 12.3 11.3–13.4 81 (15.3) 1.9 1.5–2.4
 60–69 687 (22.6) 24.2 22.5–26.1 574 (22.8) 20.3 18.6–22.0 113 (21.3) 4.0 3.3–4.8
 70–79 755 (24.8) 36.8 34.2–39.5 620 (24.7) 30.2 27.9–32.7 135 (25.5) 6.6 5.5–7.8
 ≥ 80 344 (11.3) 39.4 35.3–43.8 235 (9.3) 26.9 23.6–30.6 109 (20.6) 12.5 10.3–15.1
Females
 00–29 102 (2.8) 0.6 0.5–0.7 63 (2.7) 0.4 0.3–0.5 39 (3.0) 0.2 0.2–0.3
 30–39 170 (4.7) 2.5 2.1–2.9 126 (5.3) 1.8 1.5–2.2 44 (3.4) 0.6 0.5–0.9
 40–49 404 (11.1) 6.3 5.7–7.0 301 (12.7) 4.7 4.2–5.3 103 (8.1) 1.6 1.3–2.0
 50–59 585 (16.0) 13.3 12.2–14.4 425 (17.9) 9.6 8.7–10.6 160 (12.5) 3.6 3.1–4.2
 60–69 759 (20.8) 23.6 21.9–25.3 528 (22.3) 16.4 15.0–17.9 231 (18.1) 7.2 6.3–8.2
 70–79 916 (25.1) 32.9 30.8–35.2 586 (24.7) 21.1 19.4–22.9 330 (25.8) 11.9 10.6–13.2
 ≥ 80 714 (19.6) 40.3 37.4–43.4 342 (14.4) 19.3 17.3–21.5 372 (29.1) 21.0 18.9–23.3
  • 95% CI: 95% confidence interval.
  • a 95% CI for rate. Rates and 95% CI values are rounded.

Like cutaneous melanoma, both ocular and mucosal melanomas are more common among people classified as white compared with people classified as black (Table 4). For cutaneous melanoma, the rates among whites were 16 times higher than among blacks. Ocular melanoma rates were 8–10 times higher among whites and mucosal melanoma rates were approximately double. Mucosal melanomas account for 4.7% of all melanomas among black men and for < 1% among white men. Mucosal melanomas account for 13.4% of all melanomas among black women but only 2.1% of all melanomas among white women. Vulvar melanomas were more common among white women (519 cases, age-adjusted rate 1.3 per million, 95% CI = 1.2–1.4) compared with black women (18 cases, age-adjusted rate 0.5 per million, 95% CI = 0.3–0.8). Although there were few vaginal melanomas among black women (13 cases total), the age-adjusted rate (0.4 cases per million, 95% CI = 0.2–0.6) was comparable to white women (118 cases, age-adjusted rate 0.3 per million, 95% CI = 0.2–0.3).

Table 4. Age-Adjusted Incidence Ratesa of Melanoma by Race, Gender, and Subsite, United States, 1996–2000
Characteristics Total no. of cases Percent of all cases Rate per million 95% CI for rate White/black rate ratio 95% CI for rate ratio
Cutaneous melanoma
 Male
  White 67,211 96% 209.2 207.6–210.8
  Black 336 89% 12.7 11.3–14.3 16.5 16.0–17.0
 Female
  White 51,574 94% 137.6 136.4–138.8
  Black 326 81% 8.6 7.7–9.6 16.0 15.5–16.5
Ocular melanoma
 Male
  White 2410 3% 7.5 7.2–7.8
  Black 25 7% 0.9 0.6–1.4 8.3 7.0–10.0
 Female
  White 2266 4% 5.9 5.6–6.1
  Black 23 6% 0.6 0.4–0.9 9.8 8.7–11.1
Mucosal melanoma
 Male
  White 474 0.7% 1.6 1.4–1.7
  Black 18 4.7% 0.7 0.4–1.2 2.3 1.5–3.4
 Female
  White 1151 2.1% 2.8 2.7–3.0
  Black 54 13.4% 1.5 1.1–2.0 1.9 1.5–2.3
  • 95% CI: 95% confidence interval.
  • a Rates are age adjusted to the 2000 U.S. standard population, with estimated 95% CIs. Rates and 95% CI values are rounded.

Cutaneous melanoma rates among whites were higher both in coastal states, compared with noncoastal states, and in southern states compared with northern states (Table 5). This pattern was not observed for either ocular or mucosal melanomas. In fact, the rates of ocular melanoma were lower in southern states compared with northern states (rate ratio 0.87, 95% CI = 0.83–0.91), mainly because of a lower rate of choroid melanoma (rate ratio 0.80, 95% CI = 0.75–0.86). In contrast, however, melanomas of the iris and ciliary body were more common in southern states compared with northern states (rate ratio 1.38, 95% CI = 1.16–1.65) and in coastal states compared with noncoastal states (rate ratio 1.28, 95% CI = 1.07–1.52). The south-to-north rate ratios for nasal and oral melanomas were 1.02 (95% CI = 0.82–1.27) and 1.17 (95%, CI = 0.83–1.64), respectively. Neither was statistically significant.

Table 5. Age-Adjusted Incidence Ratesa of Melanoma by Region and Subsite, Whites Only, United States, 1996–2000
Characteristics Total no. of cases Rate per million 95% CI for rate Rate ratio 95% CI for rate ratio
Cutaneous melanoma
 Coastal 60,733 186.2 184.8–187.7 1.23 1.22–1.25
 Noncoastal 58,052 151.2 149.9–152.4
 Southern 51,761 179.5 178.0–181.1 1.13 1.12–1.14
 Northern 67,024 159.0 157.8–160.2
Ocular melanoma
 Coastal 2188 6.7 6.5–7.0 1.05 1.00–1.09
 Noncoastal 2488 6.4 6.2–6.7
 Southern 1734 6.0 5.8–6.3 0.87 0.83–0.91
 Northern 2942 6.9 6.7–7.2
Mucosal melanoma
 Coastal 750 2.3 2.2–2.5 1.05 0.92–1.18
 Noncoastal 875 2.2 2.1–2.4
 Southern 634 2.2 2.1–2.4 0.96 0.85–1.08
 Northern 991 2.3 2.2–2.4
  • 95% CI: 95% confidence interval.
  • a Rates are age adjusted to the 2000 U.S. standard population, with estimated 95% CIs. Rates and 95% CI values are rounded.

The majority (68.6%) of noncutaneous melanomas lacked specific morphologic codes. Among mucosal melanomas with specific morphology (n = 477), 28.1% of tumors were coded to nodular melanoma, 35.4% were coded to superficial spreading, and 15.7% were coded to spindle cell or mixed epithelioid and spindle cell. Among ocular melanomas with specific morphology (n = 1626), 12.4% of tumors were coded to epithelioid, 36.0% were coded to mixed spindle cell and epithelioid, 21.4% were coded to spindle cell melanoma type B, and 23.9% were coded to spindle cell melanoma with no specified type. Ocular melanomas were less likely to be microscopically confirmed (66.1%) compared with skin (99.1%) and mucosal melanomas (99.4%). Among ocular melanomas, microscopic confirmation was lowest for tumors occurring in the choroid (58.9%). According to data from the NCDB, 30% of patients with ocular melanoma are treated with radiotherapy only, which would suggest that tissue specimens would not be available for microscopic confirmation.1

Approximately 16% of both ocular and mucosal melanomas did not have stage information available for these analyses (excluding cases diagnosed in California). Among ocular melanoma cases with known stage (n = 3285), the majority (92.2%) were confined to the eye (localized summary stage). The stage distribution of mucosal melanomas varied by anatomic site. Among genital tumors with known stage (n = 543), 67.0% were localized, 23.9% were regional, and 9.0% had distant metastasis. Mucosal melanomas in other sites all tended to be diagnosed at a later stage than genital melanomas, and had similar stage distributions among both men and women. Of mucosal melanomas other than genital with known stage (n = 630), 41.3% were localized, 33.0% were regional, and 25.7% had distant metastasis. Of mucosal melanoma cases diagnosed at a regional stage and for which further summary staging information was available (n = 284), 45.1% of cases had metastases to regional lymph nodes, which has been associated with poorer survival than regional spread by direct extension only.1

DISCUSSION

These population-based noncutaneous melanoma rates are based on a larger number of these tumors than included in any previously published study. The size of the database available from the NAACCR allows for the calculation of stable rates for very rare tumors such as the mucosal melanomas. Greater than 60% of the U.S. population was covered by high-quality cancer registries that agreed to participate in the NAACCR analytic data file. Applying the age-specific rates observed in the current study to the U.S. population, an expected 2200 noncutaneous melanomas occur in the United States every year. This includes approximately 830 men and 790 women diagnosed with ocular melanoma annually and 170 men and 420 women diagnosed with mucosal melanoma annually.

Most other estimates of the incidence of noncutaneous melanoma in the United States are based on data from the SEER program, which covers 10–14% of the U.S. population.2, 9-14 All of the previous reports that included age-adjusted rates used the 1970 U.S. population standard for age adjustment. Therefore, the rates are not directly comparable to those presented in the current study. However, when our analyses were repeated with the 1970 standards, they were in general agreement with previous reports. For example, Inskip et al.2 reported an age-adjusted rate among white males and females combined of 5.7 cases per million residents, age adjusted to the 1970 U.S. population for 1994–1998 (based on 658 cases).2 The comparable rate adjusted to the same standard from the NAACCR data for the United States is 5.6 cases per million residents for 1996–2000 (based on 4676 cases).

The ocular melanoma rates reported in the current study were lower than those reported for Australia. Among white men, the age-adjusted (world standard) rate was 5.3 cases per million in the United States compared with 8.0 cases per million in Australia. Among white women, the rate was 4.2 cases per million in the United States compared with 6.1 cases per million in Australia.3 The cutaneous melanoma rates in Australia are twofold higher than in the United States.15 In Australia, Vajdic et al.3 found that 20% of ocular melanoma cases were not reported to the state or territorial cancer registries, although the authors note that the underreporting was highest for the last year of the study, and some portion of the underreporting may have been due to late reporting. This 20% underreporting rate was higher than the 8% underreporting rate that was observed in Israel.16 In both studies, underreporting was related to clinically diagnosed tumors with noninvasive treatment. Underreporting of cutaneous melanomas has been observed among population-based cancer registries in the United States, particularly among cancer registries not participating in the SEER program.17 This underreporting is most likely due to reliance on case reports from hospitals (inpatient and outpatient). According to the SEER program data, 15% of cutaneous melanomas are reported from laboratories and 12% from physician's offices.18 The lack of nonhospital reporting sources that gives rise to the underreporting of cutaneous melanoma in the United States may not present similar issues for ocular and mucosal melanomas. According to SEER data, 98% of ocular melanomas and 96% of mucosal melanomas were reported from hospital-based reporting sources. The possibility of underreporting of clinically diagnosed cases cannot be ruled out, however.19

Despite the shared cellular origin, there are some differences in the epidemiology of cutaneous and noncutaneous melanomas. Mucosal melanomas have a generally older age distribution than is observed for cutaneous melanoma, and are diagnosed more often at an advanced stage of disease.1 Familial disposition has been suggested as a risk factor for ocular melanoma.20-22 Large numbers of nevi may be risk factors for both ocular and cutaneous melanoma.23 Although melanomas are more common among whites compared with blacks in the United States, the differences are less pronounced for mucosal melanoma than for cutaneous and ocular melanomas. For mucosal melanoma, the rate among whites is 2 times higher than among blacks, whereas the rates of cutaneous and ocular melanoma are 5–20 times higher.9, 10, 24, 25 These findings suggest that cutaneous and ocular melanomas may share a common environmental risk factor (i.e., ultraviolet [UV] radiation), which may not be shared with mucosal melanomas. The evidence of an association between ocular melanoma and UV radiation is equivocal.26-35 A recent study in Australia suggested an association between UV light exposure and choroidal and ciliary body tumors, but not with tumors of the iris and conjunctiva.26 In the current study, rates of melanoma of the ciliary body and iris were higher in southern and coastal regions, but choroid melanoma rates were higher in southern states. Unfortunately, ICD-O-2 coding cannot distinguish iris from the ciliary body. Internationally, studies using place of residence, either at birth or at time of diagnosis, as a surrogate for ambient solar radiation, have not been conclusive.3, 26, 29, 30, 34, 35 For example, several studies in the United States have shown a higher risk of ocular melanoma with a southern birthplace or residence.31, 32 Several studies of cutaneous melanoma in the United States have found gradients in melanoma incidence according to latitude and residence in coastal locations,36-40 although some have suggested that this association is decreasing over time.37, 38 A previous report of north/south differences in nasal and oral melanomas was not replicated in the current study.9 A weak, nonsignificant latitude gradient was observed for ocular melanoma in Australia.3 The use of state of residence as was available for the current study is, at best, a crude measure of UV radiation exposure. Although the large population covered by the NAACCR allows more detailed examination of rates by subsite and population group, caution must be used when interpreting small differences in rates such as the marginally significant 5% excess of ocular melanoma among coastal states compared with noncoastal states.

An increasing proportion of the U.S. population is covered by high-quality, population-based cancer registries.41, 42 The availability of high-quality data from the NAACCR provides the opportunity to examine the incidence of rare tumors such as noncutaneous melanomas by demographic and geographic characteristics. The current study has confirmed and strengthened many previous findings from case series such as the NCDB report on cutaneous and noncutaneous melanoma,1 and provides a benchmark for comparison of rates and site distribution both in the United States and internationally.