Kaplan-Meier curves for overall survival according to race/ethnicity.
Kaplan-Meier curves for melanoma-specific survival according to race/ethnicity.
Stage I (A), stage II (B), stage III (C), and stage IV (D) melanoma-specific survival for African Americans vs whites.
Cormier JN, Xing Y, Ding M, Lee JE, Mansfield PF, Gershenwald JE, Ross MI, Du XL. Ethnic Differences Among Patients With Cutaneous Melanoma. Arch Intern Med. 2006;166(17):1907-1914. doi:10.1001/archinte.166.17.1907
Copyright 2006 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2006
Melanoma incidence continues to increase in whites, but little is known about melanoma in minority populations. Surveillance, Epidemiology, and End Results (SEER) data were used to examine the incidence, manifestations, and survival in patients with melanoma with respect to race/ethnicity.
A SEER search (1992-2002) for primary invasive cutaneous melanoma cases identified 48 143 whites, 932 Hispanics, 394 Asian/Pacific islanders, 251 African Americans, and 52 American Indians. Multivariate analyses were performed to evaluate the relationship between race/ethnicity and clinicopathologic factors; associations between race/ethnicity and survival were examined using the Cox proportional hazards model.
Based on our cohort of patients, the average annual age-adjusted melanoma incidence per 100 000 persons was 18.4 for whites compared with 2.3, 0.8, 1.6, and 1.0 for Hispanics, African Americans, American Indians, and Asians, respectively. Lower extremity and acral lentiginous melanomas were more common among minorities. Overall 5-year survival was 72.2% to 81.1% for minorities compared with 89.6% for whites. A 1.96- to 3.01-fold greater risk of disease-specific mortality persisted in minorities compared with whites after adjusting for age, sex, and region. In addition, Hispanics (odds ratio [OR], 3.6), African Americans (OR, 4.2), American Indians (OR, 3.4), and Asians (OR, 2.4) were more likely to present with stage IV melanoma than were whites. African Americans had a 1.48-fold higher rate of risk-adjusted, stage-specific mortality compared with whites.
Melanoma is a public health concern for all ethnic populations. Differences in disease stage at presentation contributes to disparities in survival. Understanding melanoma in minority populations may lead to early detection and ultimately save lives.
Incidence rates for melanoma have been increasing rapidly for several decades in white populations.1 A population-based study from the California Cancer Registry recently reported that rates of invasive melanoma have also increased markedly among Hispanics since 1988.2 In general, there is a paucity of literature pertaining to the incidence of melanoma in minority populations.3,4
Primary melanoma occurs more commonly in unusual anatomic sites (eg, palms and soles) in minority populations than in whites.5 Given the reported rarity of its occurrence and its unusual presentation, it is not surprising that the diagnosis of melanoma is often delayed in minority populations, resulting in more advanced stages of disease at presentation.6 This delay in diagnosis is critical because early detection and treatment of thin melanomas results in better survival. Considering that the US Census Bureau has projected that the population of Hispanics and Asians in the United States will increase by 30% from 2000 to 2010 and that the overall incidence of melanoma continues to rise,7 it is imperative that clinicians have an understanding of the manifestations and outcomes of melanoma in minority patients so that early detection and treatment is not overlooked in these populations. In this study, we examine contemporary data from Surveillance, Epidemiology, and End Results (SEER) tumor registries to examine disparities among various ethnic groups in the United States in terms of melanoma incidence, histologic subtype, disease stage at presentation, and disease-specific survival.
Data from the SEER program of the National Cancer Institute public use data set were extracted for population data and patients diagnosed as having primary cutaneous invasive melanoma from 1992 through 2002. During this time, the SEER program supported population-based prospective tumor registries in the following 11 areas: the metropolitan areas of San Francisco/Oakland, Calif; Detroit, Mich; Atlanta, Ga; and Seattle, Wash; Los Angeles County, California; the San Jose/Monterey area, Calif; and the states of Connecticut, Iowa, New Mexico, Utah, and Hawaii. The SEER data include demographic characteristics such as age, sex, race/ethnicity, and marital status; primary tumor location; pathologic data, including histologic tumor type; Breslow thickness; presence or absence of primary tumor ulceration; lymph node metastasis and distant organ metastases; and treatment rendered in the first 4 months following diagnosis.
Marital status is categorized in SEER into 3 groups: married, unmarried (single, separated, divorced, or widowed), and unknown. For this analysis, race/ethnicity was categorized as white (non-Hispanic), Hispanic, African American, American Indian, or Asian/Pacific islander. The SEER registries were combined into 4 geographic regions: West (San Francisco, Hawaii, Seattle, San Jose/Monterey, Los Angeles, Utah, and New Mexico); Midwest (Iowa and Detroit); Northeast (Connecticut); and South (Atlanta). The objective of using regions instead of individual registries in SEER was to account for potential geographic variation while still including smaller registries with fewer minority patients in the multivariate analysis. The year 1992 was selected as the beginning of the study period because in that year 2 registries were added to the SEER program—those serving the urban areas surrounding Los Angeles and San Jose/Monterey, both of which contain racially and ethnically diverse populations.
All patients with a diagnosis of primary cutaneous melanoma (identified by the SEER code 25010)8 during the years 1992 through 2002 were potentially eligible for inclusion. Patients with melanoma in situ (corresponding to summary stage 1977 SEER code 0) were excluded, thus restricting the study cohort to patients with a histologic diagnosis of invasive melanoma (histologic type, International Classification of Diseases for Oncology, Third Edition [ICD-O-3], SEER codes 8720, 8721, 8723, 8730, 8741, 8743-8746, 8770-8772, and 8780). We further selected those patients with adequate pathologic information for classification according to the American Joint Committee on Cancer (AJCC) staging system.9 Given these criteria, 53 159 patients with invasive melanoma were potentially eligible. Additional patients were excluded for the following reasons: cancer was diagnosed at autopsy or only on the death certificate (n = 118), race/ethnicity was classified as “other” (n = 69) or “unknown” (n = 2360), and Hispanic ethnicity was classified as “Spanish surname only” or “unknown” (n = 834). In addition, for the race/ethnicity categories to be mutually exclusive, 5 African Americans and 1 Asian/Pacific Islander were excluded because they were also categorized as being Hispanic.
The study cohort thus included 49 772 patients with primary invasive melanoma, including 48 143 whites (96.7%), 932 Hispanics (1.9%), 394 Asian/Pacific islanders (0.8%), 251 African Americans (0.5%), and 52 American Indians (0.1%).
The average age-adjusted melanoma incidence rates (per 100 000 persons) were calculated for each race/ethnicity group standardized by the US population in 2000. The estimated annual percentage change over 11 years (1992-2002) was calculated with 95% confidence intervals (CIs) to measure trends over time for each race/ethnicity.
Descriptive characteristics for each race/ethnicity group examined include the following: age at diagnosis, sex, marital status, SEER registry location, year of diagnosis, primary tumor location, histologic subtype, and AJCC tumor stage.9 Unadjusted probabilities for disease-specific survival were estimated using the method of Kaplan and Meier,10 and intergroup comparisons were made using the log-rank test.11
Associations between race/ethnicity and disease-specific survival were estimated using the Cox proportional hazards regression model.12 Cox regression analysis was performed using SAS version 9.1 (SAS Institute, Cary, NC) for Windows to compute hazard ratios and 95% CIs and to evaluate the effects of confounding and modifying factors. We used 2 models to assess disease-specific survival: one that was adjusted for age, sex, and SEER region and a second, more complex model that accounted for other potential confounders including marital status, year of diagnosis, primary tumor site, histologic subtype, and AJCC stage. In all of the models, whites served as the reference group for race/ethnicity, and stage I was used as the reference category for AJCC stage.
Using statistical software (Stata/SE version 8 for Windows; StataCorp, College Station, Tex), multinomial logistic regression was performed to compute odds ratios (ORs) and 95% CIs13 and to evaluate the effects of confounding and modifying factors on the association between race/ethnicity and AJCC melanoma stage. In this model, the reference category used was stage I. In addition to adjusting all analyses for age, sex, marital status, year of diagnosis, region, anatomic site, and histologic subtype, we examined AJCC stage at presentation using whites as the reference group. All P values were 2-sided and considered significant at the .05 level.
Based on the eligible cohort, the average annual age-adjusted melanoma incidence per 100 000 persons was 18.4 (95% CI, 18.3-18.6) for whites compared with 2.3 (95% CI, 2.2-2.5), 0.8 (95% CI, 0.7-0.9), 1.6 (95% CI, 1.2-2.1), and 1.0 (95% CI, 0.9-1.2) for Hispanics, African Americans, American Indians, and Asian/Pacific islanders, respectively. The estimated annual percentage change in melanoma incidence was 5.0% (95% CI, 4.1% to 5.9%) for whites compared with 1.0% (95% CI, −2.6% to 4.8%), 3.8% (95% CI, −3.4% to 11.6%), −0.6% (95% CI, −7.6% to 7.0%), and 6.1% (95% CI, 2.9% to 9.4%) for Hispanics, African Americans, American Indians, and Asian/Pacific islanders, respectively.
The clinical characteristics of the study cohort stratified according to race/ethnicity are presented in Table 1. The mean age at melanoma diagnosis was 57 years for whites, compared with 54 years for Hispanics, 59 years for African Americans, 52 years for American Indians, and 57 years for Asian/Pacific islanders (P<.001). Female sex represented a larger proportion of patients in minority populations than in the white population (50.6% to 54.5% vs 43.9%, respectively; P<.001). Compared with other ethnic populations, fewer African American patients (35.9%) were married. Overall, a majority of patients (62%) were from the West region. In particular, a large percentage of Hispanics (91.6%), American Indians (94.2%), and Asian/Pacific islanders (93.4%) were enrolled from the West.
A comparison of the clinicopathologic characteristics by race/ethnicity is summarized in Table 2. The truncal region was the most common primary tumor site for whites (34.0%) and American Indians (34.6%), whereas the lower extremity was the most common primary site for Hispanics (30.0%), African Americans (48.2%), and Asian/Pacific islanders (36.8%) (P<.001). Among the patients with histologic reports, superficial spreading was the most common histologic subtype for all groups. Acral lentiginous melanoma was more prevalent among minority populations than in whites: ORs, 5.5, 20.5, 4.1, and 11.5 for Hispanics, African Americans, American Indians, and Asian/Pacific islanders, respectively. Of whites, 66% presented with thin melanomas (≤1 mm thick) compared with 48.3% of Hispanics, 44.2% of African Americans, 57.7% of American Indians, and 48.7% of Asian/Pacific islanders (P<.001). The median tumor thickness (Breslow thickness) at presentation was 0.66 mm for whites, compared with 1.1 mm, 1.2 mm, 0.77 mm, and 1.07 mm for Hispanics, African Americans, American Indians, and Asian/Pacific islanders, respectively (P<.001). In addition, minority patients were more likely to have primary melanomas that were Clark level IV or higher, and Hispanics, African Americans, and Asian/Pacific islanders were more likely to have primary tumor ulceration (P<.001). Minority patients were more likely to have advanced stages of disease at presentation: 10.2% of Hispanics, 16.7% of African Americans, 15.4% of American Indians, and 9.6% of Asian/Pacific islanders presented with stage IV (distant metastatic disease) melanoma compared with only 3.9% of whites (P<.001).
The Kaplan-Meier curves for overall survival stratified according to race/ethnicity are presented in Figure 1. These curves demonstrate that whites had better overall outcomes than other minority populations, particularly African Americans. With a median follow-up of 4.6 years, 5-year survival rates were 69.7% (95% CI, 66.1%-73.4%), 58.2% (95% CI, 51.5%-65.7%), 69.8% (95% CI, 56.7%-85.8%), and 70.9% (95% CI, 65.6%-76.5%) for Hispanics, African Americans, American Indians, and Asian/Pacific islanders, respectively, compared with 79.3% (95% CI, 78.8%-79.7%), for whites (log-rank, P<.001). The distribution of nonmelanoma causes of death were also examined in this cohort. In the 5110 patients who died of causes other than melanoma, there was no significant difference among the various ethnic groups (data not shown).
The Kaplan-Meier curves for melanoma-specific survival according to race/ethnicity are presented in Figure 2. These curves demonstrate that whites had better overall outcomes compared with other minority populations, particularly African Americans. Five-year survival rates were 81.1% (95% CI, 78.1%-84.3%), 72.2% (95% CI, 65.5%-79.6%), 81.0% (95% CI, 69.2%-94.9%), and 80.2% (95% CI, 75.4%-85.2%) for Hispanics, African Americans, American Indians, and Asian/Pacific islanders, respectively, compared with 89.6% (95% CI, 89.3%-90.0%) for whites (log-rank, P<.001). The stage-specific Kaplan-Meier curves for melanoma-specific survival for African Americans vs whites are presented in Figure 3.
To find a possible explanation for the noted differences in survival, we examined the adjusted ORs using multinomial logistic methods for patients presenting with various stages of melanoma according to race/ethnicity (Table 3). Compared with whites, minority populations were approximately twice as likely to present with stage II or III melanoma. Of even greater concern were the significantly higher odds of minority patients presenting with stage IV melanoma compared with those of whites (ORs, 3.6, 4.2, 3.4, and 2.4 for Hispanics, African Americans, American Indians, and Asian/Pacific islanders, respectively). Other significant (P<.001) clinicopathologic factors associated with stage IV melanoma presentation were older age (OR, 1.1 per decade), male sex (OR, 1.5), unmarried status (OR, 1.7), and nodular (OR, 22.4) or acral lentiginous (OR, 12.8) histologic subtype.
Multivariate analyses demonstrated that after adjusting for age, sex, and region, minority populations had a 2.0- to 3.0-fold greater risk of disease-specific mortality compared with whites (P<.001) (Table 4). However, when the analysis was adjusted for disease stage at presentation and other potential confounders (eg, marital status, year of diagnosis, tumor site, and histologic characteristics), only African Americans were noted to have a statistically greater risk of disease-specific mortality compared with whites (hazard ratio, 1.5; 95% CI, 1.1-2.0). Of note, the CIs were large for American Indians and Asian/Pacific islanders, reflecting the small number of events in these groups.
Our results confirm that cutaneous melanoma is a rare disease in minority populations, with an average annual age-adjusted incidence of 2.3, 0.8, 1.6, and 1.0 per 100 000 persons in Hispanics, African Americans, American Indians, and Asian/Pacific islanders, respectively. An estimated annual percentage increase in melanoma incidence of 5.0% was noted for whites (P<.001) and 6.1% for Asian/Pacific islanders (P = .002). Although the disease is known to be less common in African Americans, those diagnosed as having cutaneous melanoma have been reported to have poorer overall survival compared with whites.4- 6,14,15 Our findings expand on these studies by providing data that include 5 main categories of race/ethnicity in a large, multisite, population-based setting with disease-specific outcomes. We demonstrated that the lower survival rates among minorities are in part due to their having more advanced stages of disease at presentation compared with whites.
It has become well established over the past decade that a disproportionate number of cancer deaths in the United States occur among racial/ethnic minorities, particularly African Americans.16,17 Potential explanations for these disparities in cancer outcomes include more advanced disease stage at presentation,18- 26 more aggressive tumor biological behavior,26- 28 differences in treatment,29 and socioeconomic status,30- 33 which includes lifestyle behaviors, social environment, and access to health care services.16,34- 36 To date, however, racial disparities with respect to melanoma outcomes have received limited attention.6,37- 39 All minority populations in this study, including Hispanics (OR, 3.6), African Americans (OR, 4.2), American Indians (OR, 3.4), and Asian/Pacific islanders (OR, 2.4), were more likely to have stage IV melanoma at the time of presentation than were whites. When outcomes were examined by stage, differences persisted only between African Americans and whites, with a 1.48-fold increase in risk-adjusted, stage-specific mortality for African Americans.
Previous reports of outcomes in patients with melanoma have been conflicting, with Reintgen et al40 reporting differences in stage-specific melanoma outcomes between African Americans and whites and, more recently, Hemmings et al,6 reporting no differences in outcomes in nonwhite vs white patients who were stratified by stage at initial diagnosis. Poorer cancer survival outcomes among minority populations for a variety of solid tumors, including breast,25 lung,41 esophageal,29,42 head and neck,26 and prostate cancers,19- 24 have been attributed to advanced cancer stage at presentation. Franke et al43 attributed the poorer prognosis of acral lentiginous melanomas to a delay in diagnosis and reported mean times for patient delay and physician delay of 4.8 years and 7 months, respectively. Given the atypical location of these lesions and the lack of awareness of this disease entity in minority populations, physician misdiagnosis has been reported as a common occurrence, further delaying potentially curative surgical treatment.44,45
Given the clinical heterogeneity of melanoma with respect to anatomic distribution and histologic subtypes in various race/ethnic groups, it has been postulated that there may be biological differences caused by genetically distinct types of melanoma.46 In support of this concept, shorter survival has been reported with nodular and acral lentiginous melanomas, which are known to be more common in minority populations.4 In the present analysis, superficial spreading melanoma was the predominant histologic subtype identified among all populations. Although the incidence of acral lentiginous melanoma was greater in minority populations (5%-17% vs 1% in whites), it still represents an uncommon histologic subtype overall. In addition, outcomes based on histologic subtype and stratified by race/ethnicity were similar (data not shown), and multivariate analyses, which controlled for both histologic subtype and disease stage at presentation, continued to demonstrate a significantly increased risk of death (hazard ratio, 1.5; 95% CI, 1.1-2.0) for African Americans compared with whites, indicating that histologic subtype alone was not an independent prognostic factor for outcome.
Racial disparities in the use of cancer-directed treatments, including surgery, radiation therapy, and chemotherapy, for a variety of malignancies have been well-documented; in particular, African Americans have been shown to be given less than optimal care.16,29,41,47- 51 Our group has previously reported that there were no significant differences between African Americans and whites with respect to melanoma treatment.52
Our study has a number of strengths. First, it is a population-based study, covering incident cases of melanoma in 11 SEER regions that accounted for 14% of the US population during the study period. The cases in this large cohort were pathologically confirmed and identified by the SEER registries, one of the most authoritative data sources for cancer. The SEER registries provide reliable information on tumor thickness, nodal status, surgical treatment, and long-term follow-up.15,18 These data allowed us to estimate the incidence, distribution, and outcomes attributable to an uncommon disease, namely cutaneous melanoma in minority populations. The rarity of melanoma in these populations has been previously described4- 6; however, accurate population-based estimates of the anatomic distribution, histologic subtypes, and overall outcomes have not been reported.
In interpreting the results of this study, it is important to acknowledge that there was a lack of information regarding some factors that may be associated with melanoma stage at presentation and differences in mortality rates. Specifically, data on socioeconomic status, comorbid conditions, and access to health care are not available in SEER. This is important because socioeconomic status has been implicated as the primary factor contributing to disparities in cancer outcomes among various ethnic groups in patients with breast cancer,25,53,54 prostate cancer,27,55- 58 lung cancer,41 head and neck cancer,26 and esophageal cancer.29,42 However, the results of additional reports examining the impact of socioeconomic status on cancer outcomes are mixed,27,31,42,59- 61 possibly indicating that the variety of measures used to estimate socioeconomic status are inaccurate or crude.32
On the basis of our study, we conclude that melanoma is an ever-increasing public health concern for minority populations as well as whites in the United States. A combination of tumor characteristics and possibly socioeconomic factors—and not treatment factors—likely contribute to the differences in stage at melanoma presentation and survival rates by race/ethnicity. Understanding these differences is of public health importance because increasing awareness of this disease among minority populations may provide an opportunity for early detection and ultimately save lives. Efforts to increase melanoma awareness should include educational programs developed for primary care physicians and patients that include the unique features of melanoma in minority patients to successfully provide screening for early detection in people of all ethnic backgrounds.
Correspondence: Janice N. Cormier, MD, MPH, Department of Surgical Oncology, Unit 444, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd, PO Box 301402, Houston, TX 77230-1402 (email@example.com).
Accepted for Publication: May 31, 2006.
Financial Disclosure: None reported.
Funding/Support: This study was funded by grant 5-K12-CA088084 from the Clinical Oncology Research Development Program, National Cancer Institute.
Previous Presentation: This study was presented at the Sixth World Congress on Melanoma; September 8, 2005; Vancouver, British Columbia.
Acknowledgment: We thank Dawn Chalaire, BA, for her editorial assistance and Orlean Smith, BS, for manuscript preparation.