Cutaneous melanoma is a disease of increasing clinical and economic importance in the United States1, 2 and the world.3 The risk of death from melanoma is related directly to the Breslow thickness of the primary lesion. This correlation has been documented so well that Breslow thickness at presentation can be used to predict mortality from the disease with confidence.4–6 This fact makes efforts at secondary prevention through early detection particularly important.
Data from the National Cancer Data Base of the United States indicate that, whereas 63% of patients with melanoma are diagnosed with in situ disease or early invasive melanoma, 37% have more advanced primary lesions and/or disease that has spread to regional lymph nodes or beyond.7 Despite the fact that approximately 1 in 63 male Americans and 1 in 85 female Americans will be diagnosed with melanoma during their lifetime,8 skin self-examination and skin examination by primary care practitioners is uncommon.9 Although physical examination of the skin is taught in United States medical schools, little reinforcement of this skill occurs in clinical training, and nondermatologist physicians are prepared poorly to recognize and diagnose melanoma as well as nonmelanoma skin carcinomas.10 In a recent study, skin self-examination was associated with a reduction in the incidence of melanoma as well as a reduction in the development of advanced disease among melanoma patients.11
Although reducing sun exposure is recommended and is considered to be the best way to prevent melanoma, there is considerable controversy surrounding the effectiveness of this intervention.12 Recent epidemiologic data from Australia, which has the highest incidence of melanoma in the world, suggests that public health campaigns to increase awareness of the disease and promote sun safety have resulted in a leveling off and decline in mortality in certain cohorts of the population.13
Melanoma is a disease of tremendous economic importance. Melanoma commonly affects a relatively young population, with various studies reporting a median age at diagnosis ranging from the mid-40s to the early 50s.14–16 Indeed, melanoma ranks fourth among cancers in loss of life years and income lost per death in persons ages 20–40 years.17 In addition, treatment costs are far greater for patients with advanced disease. Tsao and colleagues2 estimated that 90% of the total annual direct cost of treating melanoma is attributable to < 20% of patients (those with Stage III and IV disease).
Despite the importance of early detection, there are no large, prospective studies that address how patients come to diagnosis. This information is critical to the development of effective public health campaigns aimed at increasing the percentage of patients who present with early disease.
MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
The study population consisted of 471 patients who presented to Memorial Sloan-Kettering Cancer Center (MSKCC) for the treatment of cutaneous melanoma between July 1995 and May 1998. Patients with an unknown primary site, noncutaneous melanoma, distant metastases, or recurrent disease were excluded. Patients completed a self-administered questionnaire at their initial visit, and the following information was solicited: who detected the lesion (self, spouse, physician, or “other”), the precise anatomic location of the lesion, and whether the patient had a family history of melanoma. Demographic data also were collected, including age and gender. To obtain information on the anatomic site of the primary melanoma, patients marked the location of their lesion on a figure representing the human body. To determine the effect of anatomic site (i.e., lesion visibility) on the Breslow thickness of the disease at presentation, we grouped anatomic regions into “exposed” and “occult” categories.18 Anatomic regions that were considered “exposed” included the face, ear, forearm, hand, dorsal foot, and the anterior neck, arm, thigh, leg, and trunk. “Occult” areas included the scalp, sole of the foot, and posterior neck, arm, thigh, leg, and trunk.
Medical record review was used to obtain any missing information when appropriate. Members of the Department of Pathology at MSKCC reviewed all pathologic material. The Breslow thickness and Clark level of the primary melanoma were recorded.
To characterize the study population, we calculated descriptive statistics. Stratified analyses were conducted to assess effect modification by gender on the association between detection patterns and Breslow thickness. Univariate logistic regression analyses were conducted to examine the correlation between each factor of interest and Breslow thickness. Indicator variables were created and entered into the models for both categoric and dichotomous variables. Logistic regression modeling19 was used to model the correlation between patterns of detection and tumor thickness, adjusting for confounding factors. We included each potential confounder in the models, and the final model included those variables that affected the point estimate by at least 10% as well as those that we considered to be of a priori biologic importance.20 Odds ratios and 95% confidence intervals (95% CIs) are presented.
- Top of page
- MATERIALS AND METHODS
Characteristics of the study population are presented in Table 1. The mean age was 53.5 years, and there were approximately equal numbers of females and males. Females were younger than males (51 years vs. 55 years, respectively; P < 0.01). Most of the patients were Caucasian (n = 456 of 471 patients; 97%), and most patients presented with American Joint Committee on Cancer Stage I and Stage II disease.21 Information regarding the Breslow thickness of the melanoma was available in 454 patients. Most patients presented with melanoma > 0.75 mm in Breslow thickness (62%; n = 283 patients). The remaining patients (38%) had thin melanoma (≤ 0.75 mm; n = 122 patients) or in situ disease (n = 49 patients).
Table 1. Descriptive Statistics of Melanoma Patients (n = 471)
|Mean age (yrs) ± SD||53.5 ± 15.5|
|Median tumor thickness (mm)a||1.13 (n = 454)|
|Age groups (yrs) (%)|
| < 30||36 (7.6)|
| 30–39||66 (14.0)|
| 40–49||80 (17.0)|
| 50–59||117 (24.8)|
| 60–69||100 (21.2)|
| ≥ 70||70 (14.9)|
| Males:||251 (53.3)|
| Females:||220 (46.7)|
|Family history of melanoma (%)|
| Yes||85 (18.0)|
| No||366 (77.7)|
|Stage at diagnosis (%)|
| 0||49 (10.4)|
| I||191 (40.6)|
| II||205 (43.5)|
| III||25 (5.3)|
|Breslow tumor thickness (mm) (%)|
| ≤ 0.75||171 (36.3)|
| > 0.75–1.50||116 (24.6)|
| > 1.50–3.00||96 (20.4)|
| > 3.00||71 (15.1)|
|Anatomic site (%)b|
| Occult area||226 (48.0)|
| Exposed area||236 (50.1)|
|Person who discovered melanoma (%)c|
| Patient||270 (57.3)|
| Physician||74 (15.7)|
| Spouse||51 (10.8)|
| Other||21 (4.5)|
| Unknown||55 (11.7)|
Family history information was available for 451 patients. Of these, 84 patients (19%) reported a family history of melanoma, and 366 patients (81%) reported no first- or second-degree relative with the disease.
Most patients (416 of 471 patients; 88%) were able to recall the individual who initially was concerned about the melanoma, and, in most cases, this was the patient himself or herself (n = 270 patients; 57%). Physician detection was the second most frequent occurrence and was reported by 16% of patients (n = 74). Detection by a spouse was the third most common pattern and was reported by 11% of patients (n = 51).
Patterns of detection were influenced by patient gender (Table 2). Females were more likely to detect their own melanoma compared with males (69% vs. 47%, respectively; P < 0.0001). Despite a trend toward thinner melanoma in females, the difference in the median Breslow thickness between females and males was not significantly different (1.10 mm vs. 1.13 mm; P = 0.07). When patients reported that their spouse detected the melanoma, it usually was the wife detecting the melanoma on her husband (46 of 51 patients; 90%; P < 0.0001). Although self-detection was more common in females, physician detection was more common in males (19% vs. 12%; P < 0.04). It is interesting to note, however, that physician detection occurred in only 4 of 84 males under age 50 years compared with 43 of 166 males age ≥ 50 years (P < 0.0001). Physician detection occurred in only 12% of women and was not statistically different between the two age groups. Self-detection occurred in a similar percentage of patients whether they were age < 50 years or older.
Table 2. Characteristics of Individuals Who Discovered Melanoma, by Gender (n = 471)a
|Characteristic||Males (n = 251)||Females (n = 220)|
|No. (%)||No. (%)||No. (%)||No. (%)||No. (%)||No. (%)||No. (%)||No. (%)|
|All||251||47 (18.7)||119 (47.4)||45 (17.9)||8 (3.2)||220||27 (12.3)||151 (68.6)||6 (2.7)||13 (5.9)|
|Age groups (yrs)b|
| < 50||84||4 (4.8)||46 (54.8)||17 (20.2)||5 (6.0)||98||9 (9.2)||68 (69.4)||4 (4.1)||4 (4.1)|
| ≥ 50||166||43 (25.9)||73 (44.0)||28 (16.9)||3 (1.8)||121||18 (14.9)||83 (68.6)||2 (1.7)||8 (6.6)|
|Family history of melanoma|
| Yes||39||6 (15.4)||16 (41.0)||11 (28.2)||1 (2.6)||46||5 (10.9)||34 (73.9)||1 (2.2)||2 (4.4)|
| No||199||38 (19.1)||97 (48.7)||34 (17.1)||7 (3.5)||167||20 (12.0)||113 (67.7)||5 (3.0)||10 (6.0)|
|Stage at diagnosisc|
| 0/I||121||31 (25.6)||47 (38.8)||19 (15.7)||5 (4.1)||119||19 (16.0)||73 (61.3)||5 (4.2)||6 (5.0)|
| II/III||129||16 (12.4)||71 (55.0)||26 (20.2)||3 (2.3)||101||8 (7.9)||78 (77.2)||1 (1.0)||7 (6.9)|
|Tumor thickness (mm)d|
| ≤ 0.75||87||28 (32.2)||27 (31.0)||19 (21.8)||5 (5.8)||84||17 (20.2)||52 (61.9)||3 (3.6)||5 (6.0)|
| > 0.75–1.50||58||7 (12.1)||29 (50.0)||9 (15.5)||1 (1.7)||58||3 (5.2)||40 (69.0)||2 (3.5)||4 (6.9)|
| > 1.50–3.00||53||9 (17.0)||29 (54.7)||8 (15.1)||2 (3.8)||43||4 (9.3)||32 (74.4)||1 (2.3)||2 (4.7)|
| > 3.00||46||3 (6.5)||29 (63.0)||9 (19.6)||0 (0.0)||25||1 (4.0)||20 (80.0)||0 (0.0)||2 (8.0)|
| Occult area||139||29 (20.9)||56 (40.3)||31 (22.3)||7 (5.0)||87||12 (13.8)||57 (65.5)||6 (6.9)||6 (6.9)|
| Exposed area||109||18 (16.5)||61 (56.0)||14 (12.8)||1 (0.9)||127||14 (11.0)||90 (70.9)||0 (0.0)||7 (5.5)|
The anatomic site of the primary melanoma was recorded in 462 of 471 patients (98%). Approximately half of the patients had melanoma arising on a visually occult primary site (226 of 462 patients; 49%), and half had melanoma arising on an exposed area (236 of 462 patients; 51%). An occult primary site was more common in males (55% vs. 40%; P < 0.001). An exposed primary site was more common in females (58% vs. 43%; P = 0.002).
A logistic regression analysis was performed to examine the correlation between detection patterns and Breslow thickness ≤ 0.75 mm, adjusting for age, gender, family history of melanoma, and anatomic site of the primary lesion. These data are presented in Table 3. Physicians were 3.6 times more likely to detect thin lesions compared with self/spouse or other nonphysician detectors (95% CI, 2.1–6.5; P < 0.0001). In addition, patients who reported a family history of melanoma had a 2.7-fold increased likelihood of presenting with a thin lesion (95% CI, 1.6–4.7; P = 0.0003). There was no significant correlation between age, gender, or lesion visibility and tumor thickness.
Table 3. Multivariate Relation between Physician Detection and Breslow Tumor Thickness ≤ 0.75 mm (n = 471)a
|Predictor||Unadjusted odds ratio||Adjusted odds ratiob||95% confidence interval||P value|
|Physician detectionc||3.29||3.64||2.06, 6.54||0.0001|
|Age ≥ 50 yrsd||1.13||1.03||0.65, 1.63||0.91|
|Family history of melanomaf||2.20||2.71||1.58, 4.70||0.0003|
|Lesion visibilityg||1.34||1.41||0.91, 2.19||0.13|
- Top of page
- MATERIALS AND METHODS
Gender differences in the patterns of melanoma detection and knowledge about melanoma have been reported previously.22–24 Women are more likely to perform skin self-examination11, 23, 25, 26 and to seek skin cancer screening.27 In addition, they are more knowledgeable about the disease.25, 28–30 Koh and Rogers31 noted gender differences in patterns of detection, with females more likely to discover their own lesions as well as those of their spouse. Other investigators have reported that males are more likely to present with a thick primary melanoma.24
Melanoma is a unique tumor, because it arises on the skin, allowing the visual detection of the lesion at an early stage.7 Despite this, many patients who present with invasive disease have a relatively long history of symptoms and signs prior to seeking evaluation. Cassileth and colleagues32 have shown that patients who experienced changes in size, color, or elevation in a pigmented lesion waited an average of 1 year before presenting to a physician for evaluation. Symptoms of bleeding or ulceration resulted in a shorter delay in presentation, but these symptoms were associated with deeper lesions.32 A large proportion of patients in that study had seen a physician in the year preceding their diagnosis.32 Other investigators have reported similar findings.33
The reasons for delay in diagnosis in patients with signs and symptoms of melanoma are multifactorial, but they certainly include a lack of awareness of the disease among the public as well as physicians in the United States and a failure of patients and physicians to examine the skin routinely. Geller and colleagues9 observed that, in a cohort of patients with recently diagnosed melanoma, skin examination by their physicians in the year preceding diagnosis was uncommon. This occurred despite the fact that 63% of patients had recently seen their regular physician. Internists, family practitioners, and gynecologists are the individuals most likely to evaluate a cutaneous lesion initially, yet most have received little or no training in skin examination. Other investigators have reported that primary care physicians in the United States are unable to diagnosis both melanoma and nonmelanoma skin carcinomas with an acceptable degree of accuracy.10 Although our data demonstrate that physician detection was associated with the diagnosis of a thin lesion, it is interesting that physicians detected the melanoma in only 16% of patients in our study.
Others have reported similar patterns of detection in patients with melanoma. Koh and colleagues34 surveyed 216 patients and found that self-detection was most common (53%), followed by physician detection (26%) and detection by family members (17%). Epstein and colleagues35 reported similar findings in a study conducted at the Johns Hopkins Melanoma Center. They found that physician detection was associated with presentation with thin melanoma, yet physicians detected the melanoma in only 24% of patients. Self-detection was most common, occurring in 55% of the patients. It is noteworthy that those authors also found that a high percentage of patients (16%) reported a family history of melanoma, but they did not state whether this was associated with presentation with early disease. This is similar to the percentage of patients in our study who reported a family history of the disease (19%) and higher than would be expected based on population-based studies, in which about 2% of patients are found to have a true family history of melanoma.36, 37 Weinstock and colleagues38 have reported that patients with dysplastic nevi grossly overreport a family history of melanoma. For the purposes of this analysis, we did not validate the family history, because we were interested primarily in the influence that a patient's belief that they had a family history of melanoma had on the Breslow thickness of their melanoma at presentation. Indeed, patients who reported a family history of the disease were more likely to present with a thin, curable lesion. This suggests that awareness of the disease or a perception of increased risk of the disease, either by the patient or by a family member, is associated with presentation with early disease. This conclusion is consistent with a recent report by Oliveria and colleagues,39 who noted that patients with knowledge of the signs and symptoms of melanoma were less likely to present with a delay in diagnosis.
In the current study, lesion visibility was not associated with presentation with a thin melanoma. This is consistent with data reported by Cassileth and colleagues,32 who found no correlation between anatomic location and lesion thickness at presentation in 568 patients with melanoma. Other investigators have reported that melanoma arising on more visible body areas is more likely to be diagnosed at an early stage.18 Indeed, in our patients, melanoma arising on a visually occult site was more likely to be thinner than melanoma arising in an exposed area, although this was not statistically significant. Perhaps this is because patients are less able to be accurate about whether the lesion has changed if they cannot see it well. A history of change is one of the most accurate predictors of melanoma compared with a benign, pigmented lesion.40 Physicians may be more likely to biopsy a pigmented lesion that is located on an anatomically occult site than a lesion that is easily visible and observed by the patient.
Physician detection and a report of a family history of melanoma are associated with presentation with a thin melanoma, suggesting that awareness of the disease by physicians and the lay public is critical for early detection. Males are less likely than females to self-detect or to detect melanoma on their spouse. Public health efforts to increase melanoma awareness in males may be a particularly effective strategy for secondary prevention. Although the benefit of population screening for melanoma is an area of considerable controversy,41 targeting screening efforts for high risk groups of patients who are unlikely to self-screen may be a more cost-effective strategy. Indeed, although the incidence of melanoma rose dramatically in Australia and elsewhere during the 1980s, the mortality rates declined in females and flattened in males from 1990 to 1994.13 This suggests that media campaigns stressing the importance of skin self examination and physician recognition of melanoma have an impact on the thicker lesions and mortality and not only on increasing early detection.42