Distribution of nonwhite patients with melanoma by ethnic origin.
Survival rates of white and nonwhite patients with melanoma. Nonwhite patients have a significantly worse survival rate than white patients (P<.003).
Hemmings DE, Johnson DS, Tominaga GT, Wong JH. Cutaneous Melanoma in a Multiethnic PopulationIs This a Different Disease?. Arch Surg. 2004;139(9):968-973. doi:10.1001/archsurg.139.9.968
Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004
Cutaneous melanoma in nonwhite persons has a manifestation and a prognosis that are different than those of cutaneous melanoma in white persons.
Tertiary care university-affiliated community medical center located in a multiethnic state in which white persons are a minority of the population.
Consecutive series of 357 patients with melanoma seen between January 1994 and August 2003.
Main Outcome Measures
Ethnicity, age, sex, primary site, tumor thickness, nodal status, stage at diagnosis, and survival.
There were 208 men and 149 women who ranged in age from 15 to 93 years (mean, 58 years). Twenty-two patients initially had unknown primary sites. Of these 357 patients, 67 (18.7%) were nonwhite. There was no statistically significant difference in the age (P = .10) or sex (P = .57) distribution of these 2 populations. Nonwhite patients at initial diagnosis had thicker tumors (P = .002), more frequently had ulcerated primary tumors (P<.001), more frequently had positive nodes (P = .004), and were at a more advanced stage (P = .002) than their white counterparts. The anatomic distribution between the 2 populations was significantly different (P<.001), with a high incidence of melanoma on the sole and subungual locations and a substantially less frequent occurrence on the head and neck, trunk, and extremities in the nonwhite population when compared with the white population. The overall survival rate of the nonwhite patients was significantly worse than that of the white patients, but when stratified by stage at initial diagnosis, there was no difference in outcome.
In a multiethnic population, cutaneous melanoma in nonwhite persons is unusual but not rare. Although the diagnoses are distinctly different at initial examinations, suggesting a potential biological component, stage-for-stage outcomes are similar between white and nonwhite persons and suggest a need for early diagnostic interventions with unusual pigmented lesions in nonwhite persons.
Cutaneous melanoma is typically a disease that occurs in individuals with blue/green eyes, who have blond or red hair and fair to light complexions and freckles, and who are sun sensitive and tan poorly. According to Surveillance Epidemiology and End Results data, the age-adjusted incidences of cutaneous melanoma is 20.6 per 100 000 white persons and only 1.1 per 100 000 African American persons. Virtually all of the known risks for melanoma are related to exposure and susceptibility to ultraviolet radiation from the sun.1
When diagnosed early, cutaneous melanoma is largely curable with a simple wide surgical excision. As a result, efforts to enhance early diagnosis by educational programs represent one of the primary strategies to improve outcomes. However, the majority of these programs target individuals who are considered to be at high risk for developing the disease, ie, fair-skinned, blue-eyed individuals,2 as well as primary care physicians in an effort to raise their index of suspicion.3 Cutaneous melanoma in nonwhite persons is a relatively unusual occurrence, and the impact of educational programs among this population is unknown, particularly in an increasingly ethnically diverse population.
The state of Hawaii has a population of approximately 1.2 million people. According to the 2000 census, approximately 24.3% of these individuals are white, 1.8% African American, 41.6% Asian American, and 9.4% Hawaiian or Pacific Islander. Approximately 21% of Hawaiian individuals are of 2 or more races, with nonwhite alone or in combination with other races comprising 60.7% of the population. As a result of the geographic location of Hawaii, the multiethnic population of the state has constant exposure to the equatorial sun. The incidence of cutaneous melanoma in the white population of Hawaii (45.6/100 000) represents one of the highest age-adjusted incidences of cutaneous melanoma, more than twice that observed in the 8 Surveillance Epidemiology and End Results registries. Hawaii also has significant numbers of cutaneous melanoma in nonwhite persons.
It is generally believed that melanoma arising in nonwhite persons occurs in unusual anatomic sites and may be initially diagnosed at a more advanced stage. For these reasons, melanoma in nonwhite persons is thought to have a poorer prognosis than melanoma in their white counterparts.4 However, it is unclear whether this apparently worse prognosis represents a delay in diagnosis, a biologically distinct entity, or a combination of factors. To address this issue, we reviewed our experience in the management of cutaneous melanoma in our nonwhite patients.
Between January 1994 and August 2003, 357 patients with the diagnosis of melanoma were seen by the senior author (J.H.W). Patients were designated as white or nonwhite based on their self-designation. Nonwhite included Asian (Japanese, Chinese, and Korean), Filipino, Hawaiian and Pacific Islander, and Southeast Asian persons, individuals from combinations of these ethnic groups, and African American and Hispanic persons. Individuals of mixed racial background in which the white component represented less than 50% of their heritage were considered to be nonwhite.
Patients were surgically treated in a standardized fashion with a wide excision of the primary site, according to previously defined recommendations,5 along with immediate sentinel node staging6,7 if the tumor was equal to or greater than 1 mm in thickness or had histologic evidence of regression. Individuals with subungual melanomas underwent a distal amputation of the digit as well as sentinel node staging. Patients who either were clinically node positive at the time of initial diagnosis or had a positive sentinel node by hematoxylin-eosin staining or immunohistochemical staining underwent a formal regional lymphadenectomy. It has been our standard to recommend adjuvant interferon alfa-2b for node-positive individuals8 and biochemotherapy or dacarbazine-based chemotherapy for patients with advanced disease.9
The method of Kaplan and Meier10 was used to calculate survival curves, and significant differences were determined by the log-rank test. Using the Cox regression model, multivariate analyses of prognostic variables were performed. T test and χ2 analysis were used where appropriate. Analysis was performed with the NCSS statistical system for Windows (Kaysville, Utah).
Table 1 summarizes the characteristics of this patient population. Melanoma occurred more frequently in men than in women (men:women ratio, 1.4:1). The trunk and extremities were the most common sites of primary tumors, with 65% of patients having primary tumors at these sites. Six percent of our patients were initially diagnosed as having metastatic disease from an unknown primary site. All of these patients had metastatic disease in a regional lymph node basin (stage IIIB or IIIC). Ulcerated tumors were substantially thicker than nonulcerated tumors (ulcerated, 3.55 mm; nonulcerated, 1.45 mm; P<.001) and far more likely to have nodal metastases (P = .005).
Sixty-seven (18.7%) of the 357 patients were nonwhite. Table 2 summarizes the clinical characteristics of the white and nonwhite patient populations. There was no statistical difference in the male-to-female ratio or the age of the patient population, although nonwhite patients were slightly older than white patients at diagnosis. Asian American persons (37%) represented the largest group of nonwhite patients with melanoma, followed by Filipino persons. Seven patients (9%) were of Hawaiian or Pacific Islander ancestry (Figure 1). Nonwhite patients were significantly more likely than white patients to have thicker primary tumors (2.31 mm vs 1.45 mm; P = .002) that were ulcerated (P<.001) and to have a more advanced disease (P = .001). The median follow-up period for the entire group was 20.3 months (range, 1-241 months).
The distribution of the primary site of the primary tumor was significantly different between white and nonwhite patients (P<.001) (Table 3). Eighty-four percent of white patients had tumors arising on the skin of the head and neck, trunk, or extremities. In contrast, only 58% of nonwhite patients had tumors arising in these sites. However, 33% of nonwhite patients had tumors arising on the palms, soles, or subungual sites, whereas only 6% of white patients had tumors arising on these anatomic sites.
When compared with their white counterparts, nonwhite persons have a substantially worse prognosis (Figure 2). Cox regression analysis indicated that age (P = .40) and thickness (P = .02) were significantly related to outcome, whereas ulceration (P = .07) and ethnicity (P = .12) did not independently predict outcome. However, when stratified by stage at diagnosis (stage I or II and stage III), there was no statistically significant difference in outcome (stage I or II, P = .18; stage III, P = .26).
The typical patient with melanoma has a characteristic phenotype that includes a fair complexion, a tendency to sunburn rather than tan, blond or red hair, and blue eyes. Melanoma arising in individuals with pigmented skin is an unusual occurrence. For this reason, few data are available on the clinical and prognostic characteristics of melanoma in this population of individuals. However, there is a broadly held perception that melanoma arising in nonwhite persons not only occurs in unusual anatomic sites but is associated with a particularly grave prognosis. Our report provides further evidence that melanoma arising in nonwhite persons does indeed occur more frequently in unusual anatomic sites. However, although nonwhite patients did significantly poorer than white patients, when stratified by stage at diagnosis, melanoma in nonwhite patients has a similar prognosis as melanoma in their white counterparts.
Cutaneous melanoma in nonwhite persons is frequently manifested differently than in white persons. The predominantly volar and subungual distribution of melanoma in Hispanic, Asian American, and African American persons is well recognized. Cress and Holly4 reviewed data from the California Cancer Registry and identified a pronounced difference in the distribution of the primary site with a predilection to the lower extremity. Bergfelt et al11 found a similar predisposition for the lower extremity in Hispanic and African American persons and postulated that this might be genetically controlled. Others have noted a higher incidence of acral lentiginous melanoma in nonwhite persons.12- 14 The Japanese Melanoma Research Group examined malignant melanoma in Japan between 1987 and 1996 and found that the most common primary site was the sole of the foot, comprising 28% of cases. Subungual primary sites were also frequently observed in the Japanese population, accounting for approximately 10% of newly diagnosed melanomas.15 These findings are consistent with results from our nonwhite population, in which we observed a marked propensity for the development of these lesions in anatomic sites that are not commonly observed in the white population.
The prognostic significance of the anatomic site of origin remains controversial. Data from the University of Alabama and the Sidney Melanoma Unit demonstrated a poorer survival rate for patients with head and neck lesions when compared with patients with trunk and extremity lesions.16 Others have suggested that patients with primary tumors arising on the upper back, the upper arm, the neck, and the scalp have a worse prognosis than patients with tumors on other anatomic sites,17 but this suggestion remains controversial.18,19 Many consider tumors that arise on the palms, soles, and subungual locations to be particularly ominous.20,21 However, Wells et al22 suggested that the differences in outcome were largely explained by differences in tumor thickness at the time of diagnosis.
Other clinical and pathological variables, including age, sex, tumor thickness, ulceration, and stage at initial diagnosis, have been examined to determine their prognostic significance for cutaneous melanoma. Tumor thickness and ulceration have emerged as uniformly the most significant predictors of outcome in node-negative cutaneous melanoma.23 Our nonwhite patients were initially diagnosed as having significantly thicker primary lesions along with more frequent ulceration than their white counterparts; this explains, at least in part, the more advanced stage at initial diagnosis and the poorer outcome of these individuals. Although significantly thicker and more frequently ulcerated tumors might simply represent a delay in diagnosis, as yet unrecognized biological factors may contribute to the poorer outcome. The biological correlate of ulceration remains unclear, but tumors with ulceration tend to invade the epidermis rather than displacing normal cells. Ulceration may correspond to an increased rate of tumor proliferation and doubling. Additionally, cytogenetics studies have demonstrated a number of chromosomal abnormalities in cutaneous melanoma. Although many of these abnormalities are duplicated in uveal melanoma, losses of chromosome 3 and duplications of 8q, which are commonly observed in uveal melanoma, are relatively rare in cutaneous melanoma24 and raise the possibility of a distinct biology for different primary sites that could contribute to different outcomes.
The importance of early diagnosis cannot be overemphasized. The accessibility of the skin to direct visual examination provides a unique but challenging opportunity to diagnose melanoma early when it is most likely to be cured by a simple wide surgical excision.23 Early diagnosis of melanoma, however, is made difficult by a number of patient-related factors, including a lack of knowledge and awareness, fear, and denial,25,26 as well as the difficulty of clinical diagnosis even among experts.27,28
Franke et al29 attribute the poorer prognosis of plantar melanoma to the delay in diagnosis. Of the 68 cases of plantar melanoma reviewed, the mean times for patient delay and physician delay were 4.8 years and 7 months, respectively. The higher incidence of plantar melanomas among nonwhite persons may represent a significant obstacle to early diagnosis. Attention to volar and subungual lesions is often delayed by patients because of the atypical anatomic location and decreased visibility of such lesions.30 This delay in seeking medical attention may be further compounded by physician misdiagnosis, leading to an even greater lag time between accurate diagnosis and treatment.30,31
Early detection programs can impact the outcome of this disease. As a result, recommendations for increased educational efforts have been made in an effort to increase the early and accurate diagnosis of melanoma.25 These efforts have largely been directed at the white population. It is reasonable to presume that similar efforts at early diagnosis need to be directed at less high-risk populations, focusing on these populations' anatomically distinct distribution of primary sites. The Japanese population has been successful in implementing awareness and screening campaigns targeting specific anatomic location and the importance of early detection and treatment. Kato et al32 have observed improved survival rates for patients with acral melanoma during the last 22 years and attribute this improvement to efforts by Japanese dermatologists to educate the general public.
In the United States, with the changing demographics and the increasing number of nonwhite persons, who currently comprise approximately 30% of the country's population, and in states such as Hawaii in which the white population represents a minority of the total citizenry, the need for education and early recognition of melanoma in sites not typically associated with cutaneous melanoma in white persons may become increasingly important. The experience in Japan suggests that educational efforts need to be directed at both the general population of nonwhite groups, particularly Asian American, Hispanic, and African American persons, and medical personnel to increase awareness of melanoma among nonwhite persons and the propensity for melanoma to occur in less common anatomic locations.
Melanoma arising in nonwhite persons represents a relatively unusual although not rare clinical entity. The distinct anatomic distribution along with the markedly increased incidence of tumor ulceration suggest that the poorer outcome of melanoma in nonwhite persons may be due, at least in part, to an inherently biologically different disease. However, our report suggests that the advanced stage at the time of initial diagnosis accounts for the grave prognosis in these individuals, and our report supports continued public and primary care awareness programs to improve any diagnostic delays.
Correspondence: Jan H. Wong, MD, Department of Surgery, John A. Burns School of Medicine, University of Hawaii at Manoa, 1356 Lusitana St, 6th Floor, Honolulu, HI 96813 (email@example.com).
Accepted for publication May 19, 2004.
This study was supported in part by the Surgical Oncology Research Fund at the University of Hawaii Foundation, Honolulu.
This paper was presented at the 75th Annual Meeting of the Pacific Coast Surgical Association, Wailea, Hawaii; February 17, 2003; and is published after peer review and revision. The discussions that follow this article are based on the originally submitted manuscript and not the revised manuscript.
Stanley P. L. Leong, MD, San Francisco, Calif: While it is well known that malignant melanoma is most prevalent in the white population, it may also be present in the nonwhite population. In almost all the melanoma centers in the mainland of the United States, the overwhelming majority ethnic population is white. Therefore, it is difficult to study the characteristics of melanoma of the nonwhite populations. As pointed out in this manuscript, the distribution of ethnic populations in Hawaii is unique, with the Asian population being the majority with 42%, followed by whites with 24%, 9% Pacific Islanders, and about 2% African Americans. Therefore, the authors have captured this unique population distribution to evaluate melanoma in both the white and nonwhite populations within the same environmental conditions, particularly with respect to sun or ultraviolet exposure.
Within the last decade, 357 patients with melanoma were seen consecutively by the senior author, Dr Wong. About 19% were nonwhite, mainly of Asian descent. Based on the ethnic and clinical features, statistical analysis showed that there was no significant difference in the age distribution or sex of these 2 populations. The authors have found that nonwhites presented with thicker tumors, had more frequently ulcerated primary tumors, more frequently with positive nodes, and presented at a more advanced stage than their white cohorts, with impressive P values. Further, the anatomical distribution between the 2 populations was significantly different, with a predominance of melanoma involving the sole and subungual locations in the nonwhite group, whereas melanoma of the head and neck, trunk, and extremities was substantially less frequent in this group as compared with the white population. Based on other studies as well as the findings in this study, the overall survival of nonwhites was significantly worse than that of whites.
Thus, the authors asked the important question whether this apparent worse prognosis was due to a delay in diagnosis, a biologically distinct entity, or a combination of factors. When stratified by stage at presentation in the 2 populations, the authors found that there was no difference in outcome for both. Thus, they have established the fact that when melanoma in nonwhites is found early, the prognosis is good, rather than the broadly held perception that melanoma in nonwhites carries a grave prognosis. Thus, early diagnosis in both whites and nonwhites is critical, as potential cure can be achieved when early melanoma is resected. This information is important for both the physicians and the public at large: any changing skin lesion particularly pigmented should be evaluated as soon as possible to make sure it is not melanoma.
I would like to ask several questions:
(1) In your nonwhite group mainly of Asian descent, were they all born and raised in Hawaii?
(2) Do you have family lineage data to show if there were any mixed marriages with potential penetration of the gene pool from whites?
(3) Have you obtained history to trace the onset of the melanoma of the nonwhites, which may lead to why there was a delay in diagnosis in this group of patients resulting in a poor prognosis? If there was indeed a delay, was that due to patient's denial or physician's misdiagnosis?
(4) Why do you think that nonwhites have a propensity of developing melanoma in the sole and subungual locations as shown in this study and others as well?
(5) Can you speculate what are the basic genetic differences between these ethnic groups that may account for the different frequency and anatomical distribution of the primary melanoma?
Theodore X. O'Connell, MD, Los Angeles, Calif: I have a question regarding the conclusion that there is no difference in biology between the white and nonwhite melanomas, only delay in diagnosis. Obviously, there is a difference in overall survival, a difference in the percentage of subungual, plantar, and palmar melanoma, all of which have worse prognosis and reflect worse biology. When you claim that when you stratified stage-for-stage there is no difference, there could be a type II error in that you don't have enough patients to prove it statistically. There is certainly a trend in the 0.2 range, which means there is an 80% chance of the difference being real and only a 20% chance of its being a random error. With more patients, this may be able to be proven statistically. I would like for you to comment on the possible type II error.
Margarette Lee, MD, Honolulu, Hawaii: I am also from Hawaii, except that I practiced in a military hospital, so my patient population is more similar to the mainland. In addition to Dr Leong's good questions, I was also wondering about other risk factors. In white patients, we know that those who have so-called dysplastic nevus syndrome have a higher risk of melanoma. I have not heard about this in the nonwhite population, so maybe the authors of this paper can shed some light.
John T. Vetto, MD, Portland, Ore: I accept the basic idea that melanoma in minorities is probably stage-for-stage the same disease with the same outcome as in whites. However, as your data points out, there were some primary biological differences noted in the nonwhites, mostly the presence of more ulceration and more acral lentiginous, palmar, and sole lesions. To me, that actually argues that melanoma in these patients might be a different disease to some extent—so that is my first question: would you agree?
My second question focuses on the practical application of this data. Do the authors think that, based on this information, they might be able to identify nonwhite populations/communities that might be at higher risk and then do some kind of screening intervention in those individuals?
Jeffrey M. Pearl, MD, San Francisco: We have just heard a number of papers about the lethal nature of this disease, even in thin melanomas, and this paper has told us that it is a ubiquitous problem. We know that the risk of developing melanoma is the ultraviolet exposure in the first 18 years of life, and I am wondering what this group in particular and what the College also are doing to preach prevention and to help prevention of this disease.
Nathalie M. Johnson, MD, Portland: I grew up in the Virgin Islands and don't remember a lot about melanoma. However, I do remember one very famous person died of metastatic melanoma and his primary was subungual under his great toe. I don't know if the authors have enough information or numbers, but I was curious in your African American and Hispanic populations, were those patients solely subungual and sole vs trunk compared with the Asian population? I would find that interesting as well.
Dr Wong: Those of you who have had the opportunity to explore Hawaii may have been struck by the multiethnic composition of our state. As Dr Hemmings pointed out, whites represent a distinct minority in our state, and although our incidence of melanoma in whites is higher than anyplace else in the United States, because of that relatively small number of whites in Hawaii, our absolute numbers are really relatively small. But we have been struck by the number of nonwhites who have had this disease, and this led us to try to examine this issue and to try to address some of the long-standing issues regarding the diagnostic and biological issue in this group of patients. Let me try to answer some of the questions that were posed to us.
Dr Leong asked me about the Asian population and whether or not they were born and raised in Hawaii. Almost all of these individuals who we have seen were born and raised in this state. I have not had the opportunity to see individuals of Asian extraction who were born and raised outside of Hawaii, so these are all individuals who have lived their entire lives here in Hawaii.
It's difficult for us to break down the actual ethnic makeup of our mixed population. We truly do have a melting pot population. It's not uncommon to have in the ethnic category in our registration forms individuals who cite themselves as Hawaiian/Portuguese/Japanese/etc, so it is hard to know exactly what gene pool we are getting into with those types of individuals.
How about a delay in diagnosis? Is that a patient problem or a physician problem? My impression is that this is a patient problem. It is common when I obtain a history from these individuals that they have noticed a lesion that they just didn't think much about and only when it started to bleed or cause some symptoms do they consider that something ought to be examined. Typically, when these patients see a dermatologist, the dermatologists are quite quick to react, but not always. In summary, I would characterize it as typically a patient problem as opposed to a physician problem, although there is evidence that both contribute to the poorer prognosis.
I have no answer as to why subungual or sole lesions in fact are more common in these individuals. These are data that have been reproduced in all of the other relatively small reports regarding this disease. Whether or not this represents a biologically different entity, I cannot tell you.
Dr O'Connell, I am glad that I have to again address a type II error, and you are correct. We may have a type II error. The problem is, we don't have very many patients and so we can only do what we can do.
Dr Lee, all of our patients did not have any other characteristics of a typical dysplastic nevus syndrome. Whether or not it occurs in these individuals, I am not aware, but they did not have the classic, large, irregular, multiple pigmented lesions over their skin.
Dr Vetto, is this biology? There is some evidence, particularly with cutaneous melanoma and uveal melanomas, that there are cytogenetic differences between them. Could they extend to other sites? We don't know. We believe that ulceration may in fact represent a biological phenomenon, but at this time this is poorly defined. I would agree that this is not necessarily totally a delay in diagnosis but in fact may represent as yet an unrecognized biological factor as well.
Dr Pearl asked about possible prevention strategies. I have discussed this data with our dermatologists. Education regarding the occurrence of this entity by our dermatologists and hopefully getting them to become more aware of not only subungual and melanomas of the palms and soles but also of the trunk and extremities in their dermatological societies hopefully will start to propagate the concept that this is not that unusual a disease.
Dr Johnson, we only had 3 patients who were African American or Hispanic. All 3 did in fact have lesions on the soles. Whether that is enough to say anything, I am not sure, but it is quite striking.