Fax: (212) 794-5847
Symptoms and signs of primary melanoma
Important indicators of Breslow depth
Article first published online: 16 JUN 2003
Copyright © 2003 American Cancer Society
Volume 98, Issue 2, pages 344–348, 15 July 2003
How to Cite
Negin, B. P., Riedel, E., Oliveria, S. A., Berwick, M., Coit, D. G. and Brady, M. S. (2003), Symptoms and signs of primary melanoma. Cancer, 98: 344–348. doi: 10.1002/cncr.11513
- Issue published online: 2 JUL 2003
- Article first published online: 16 JUN 2003
- Manuscript Revised: 8 APR 2003
- Manuscript Accepted: 8 APR 2003
- Manuscript Received: 21 JAN 2003
- Breslow depth;
Operative management of patients with cutaneous melanoma is guided primarily by the pathologic determination of Breslow depth. Differentiating early from more advanced melanoma is not always straightforward and may be complicated by pathologic misdiagnosis, inappropriate biopsy techniques, or poor specimen handling. Inconsistencies between the patient's history and the pathologist's interpretation may alert the physician to the possibility of misdiagnosis. In this setting, awareness of the signs and symptoms (S/S) of thin versus intermediate or deep melanoma may be helpful in guiding management. The authors performed a prospective evaluation of the S/S reported by patients who presented at Memorial Sloan-Kettering Cancer Center with invasive primary melanoma.
The authors prospectively evaluated 369 patients with a detailed questionnaire regarding their S/S at the time of their initial visit. Multivariate logistic regression was employed to study the association between S/S reported by the patient and Breslow depth of the primary lesion, adjusting for gender, age, and anatomic site. Patients were grouped by the Breslow depth of their primary tumors into three categories for analysis: those with thin (≤ 1.0 mm), intermediate (1.0–4.0 mm), and thick (≥ 4.0 mm) lesions.
Gender, age, and primary site were not significantly predictive of increasing category of Breslow depth. Most patients reported at least one S/S (n = 278 [75%]). The most common S/S reported was an increase in size (n = 187 [51%]), followed by a change in color (n = 147 [40%]). Bleeding (n = 95 [26%]), lump (n = 86 [23%]), itching (n = 83 [22%]), skin breakdown (n = 66 [18%]), and pain (n = 24 [7%]) were less common. In a multivariate analysis, the S/S most strongly associated with an increased category of Breslow depth was bleeding (odds ratio [OR] 7.5), followed by pain (OR 3.3), lump (OR 2.2), itching (OR 1.9), and change in size (OR 1.7). The only S/Ss not independently associated with an increasing category of Breslow depth were a change in color and skin breakdown. The presence of one or more S/S was associated significantly with an increased category of Breslow depth of the primary melanoma (1 or 2 S/S vs. no S/S: OR, 4; ≥ 3 S/S vs. no S/S: OR, 24).
Most S/Ss of cutaneous melanoma are associated with an increasing risk of a deep primary lesion. Understanding this relationship can be valuable in patient management, especially when pathologic data are incomplete or inconsistent with the patient's history. Cancer 2003;98:344–8. © 2003 American Cancer Society.
The incidence of cutaneous melanoma has been increasing rapidly, reaching epidemic proportions in the United States1, 2 and around the world.3–5 In the United States, the lifetime risk for developing melanoma is currently 1 in 60 for men and 1 in 80 for women.6 This represents an increase of greater than 1800% since the 1930s.7
The prognosis for patients with cutaneous melanoma is related primarily to the Breslow depth of the primary lesion.8–10 Surgical excision with 1-cm margins is highly effective for patients with early melanoma (< 1.0 mm Breslow depth), resulting in a 93% 5-year survival rate.11 Deeper lesions require excision with wider margins (≥ 2 cm) and evaluation of regional lymph nodes. Once melanoma has spread beyond the primary lesion, the 5-year survival rate decreases, with a 10–50% survival rate for patients with locoregional recurrence (Stage III) and survival rate of less than 10% for patients with distant (Stage IV) disease.12
It is unfortunate that many surgeons have limited experience with patients who present with cutaneous melanoma and very little understanding of the signs and symptoms (S/Ss) of the disease. Dermatologists diagnose and treat many patients with early disease, but they refer patients for surgical excision when the melanoma requires a wide excision and/or evaluation of regional lymph nodes. The surgeon performs the appropriate operation based on the pathologist's report of the primary lesion's Breslow depth and Clark level. Regrettably, pathologic misdiagnosis may occur. Discordant interpretations of pigmented lesions and melanoma are not uncommon, even among expert pathologists.13 Accurate lesion depth may be difficult or impossible to determine for other reasons, including poor specimen handling or processing, inappropriate biopsy technique, or inexperienced pathologists. In addition, pathology reports frequently lack important prognostic information such as ulceration, Clark level, and even Breslow depth.14 An understanding of the relationship between the Breslow depth of the primary tumor and the S/Ss associated with early versus advanced melanoma may help to facilitate accurate diagnosis and appropriate management. Surgeons and dermatologists aware of S/Ss of early versus more advanced melanoma will be more likely to question pathology reports that appear inconsistent with the patient's history. This is particularly important for the dermatologist, who typically performs the initial assessment. To understand the relationship between lesion depth and specific S/Ss, we prospectively evaluated patients with newly diagnosed invasive melanoma presenting for surgical management.
MATERIALS AND METHODS
From May 1995 to March 1999, we identified patients presenting with newly diagnosed cutaneous melanoma and provided them with a questionnaire to complete at their initial visit. Of the 593 patients with localized invasive cutaneous melanoma who presented for treatment, 379 were given the questionnaire. There was a completion rate of 97% (n = 369). Participation in the study varied among surgical attendings, and no active selection of patients was performed. In addition, patients who were treated elsewhere and presented for a second opinion were excluded. This was done to eliminate patients for whom there was a significant delay between the diagnosis of invasive melanoma and presentation for treatment to the Memorial Sloan-Kettering Cancer Center (MSKCC). Patients were given the questionnaire before seeing the physician on the day of their initial office visit.
Patients were asked about the following S/Ss of primary cutaneous melanoma before biopsy and answered yes or no: 1) has your lesion changed in size? 2) has your lesion changed in color? 3) has your melanoma bled? 4) does it itch? 5) has the skin over it broken down? 6) is it painful? 7) did you notice a lump where the melanoma is? and 8) have you had no symptoms at all?
The anatomic location of the lesion was determined by having the patient indicate the lesion site on a figure of the human body. Incomplete data on age, gender, or anatomic site of the lesion were completed by medical record review. All data regarding S/Ss associated with the primary lesion were obtained using only the patient's answers to the questionnaire.
Pathologists at MSKCC reviewed all pathologic material pertinent to the study. Patients for whom the Breslow depth could not be determined were excluded from the study.
The primary objective of the study was to investigate whether the S/Ss reported by the patient (change in size, change in color, bleeding, itch, skin breakdown, pain, and lump) were predictive of Breslow depth. Patients were categorized into three groups (specifically, Breslow depth ≤ 1 mm; > 1 mm and < 4 mm; and ≥ 4 mm). In addition to S/Ss, we adjusted for gender, age (categorized as < 60 years old or ≥ 60 years old), and anatomic site (extremity, head/neck, or trunk). In the univariate analysis, the association between each variable and Breslow depth was assessed using the Mantel–Haenszel test for trend. Variables significant in the univariate analysis (at the 0.05 level) were entered into a stepwise logistic regression modeling procedure (using the cumulative logit model15) to determine which S/Ss were independent predictors for increasing Breslow depth category. The odds ratios (ORs) from the cumulative logit model are interpreted as the odds of being in the higher category (thicker lesions) versus the odds of being in the lower category, regardless of which adjacent depth categories are compared. The cumulative logit model assumes that the effect of a particular variable (or S/S) on the odds of having a thick lesion versus the odds of having an intermediate lesion is the same as the effect of that variable on the odds of having an intermediate lesion versus the odds of having a thin lesion. The assumption of proportional odds was checked using a score test. Based on the results of the multivariable model, patients then were grouped by the number of S/Ss with which they presented. ORs and 95% confidence intervals (CIs) were calculated. All analyses were done using SAS, Version 8.0 (SAS Institute, Cary, NC). P values of less than 0.05 were considered statistically significant.
Complete clinical and pathologic information was available for all patients. There were slightly fewer females than males (45% vs. 55%). The majority of patients had Stage I (48%) or Stage II (45%) disease, as defined by the American Joint Committee on Cancer staging criteria.8, 16 The median Breslow depth of the primary melanoma was 1.2 mm (range 0.2–17 mm). Most patients presented with thin (n = 164 [44%]) or intermediate-thickness (n = 176 [48%]) melanoma, and only 29 (8%) patients presented with thick melanoma. These and other descriptive statistics of the study population are summarized in Table 1.
|Median (range)||55 (16–91 yrs)|
|< 60||216 (59)|
|≥ 60||153 (41)|
|Median (range)||1.2 (0.2–17 mm)|
|≤ 1||164 (44)|
|> 1, < 4||176 (48)|
|≥ 4||29 (8)|
Most patients presented with at least 1 symptom or sign (n = 278 [75%]). The most common S/S experienced was an increase in size (n = 187 [51%]) followed by a change in color (n = 147 [40%]). Many patients experienced bleeding (n = 95 [26%]), described a lump at the site of the primary tumor (n = 86 [23%]), or experienced itching (n = 83 [22%]) and/or a breakdown of skin over the lesion (n = 66 [18%]). Few patients experienced pain from the melanoma (n = 24 [7%]). The frequency of specific S/Ss reported at presentation is displayed in Table 2.
|Change in size||187 (51)|
|Change in color||147 (40)|
|Skin breakdown||66 (18)|
|Any symptom||278 (75)|
Results of the univariate analysis are provided in Table 3. There was no significant association between Breslow depth and gender, age, or primary site. In addition, reporting a change in color was not associated with the depth of the primary melanoma. All of the other S/Ss were significantly associated with the increasing depth of the primary lesion.
|Factor||Breslow depth (mm)||P value|
|≤ 1 (%)||> 1 and < 4 (%)||≥ 4 (%)|
|Male||93 (46)||94 (46)||16 (8)||0.64|
|Female||71 (44)||82 (49)||13 (8)|
|< 60||64 (42)||74 (48)||15 (10)||0.24|
|≥ 60||100 (46)||102 (47)||14 (7)|
|Extremity||69 (39)||92 (53)||14 (8)||0.23|
|Head/neck||16 (67)||5 (21)||3 (13)|
|Trunk||79 (47)||79 (47)||12 (7)|
|Yes||63 (34)||103 (55)||21 (11)||< 0.0001a|
|No||101 (56)||73 (40)||8 (4)|
|Yes||63 (43)||74 (50)||10 (7)||0.89|
|No||101 (46)||102 (46)||19 (9)|
|Yes||9 (10)||66 (70)||20 (21)||< 0.0001a|
|No||155 (57)||110 (40)||9 (3)|
|Yes||20 (24)||49 (59)||14 (17)||< 0.0001a|
|No||144 (50)||127 (44)||15 (5)|
|Yes||12 (18)||43 (65)||11 (17)||< 0.0001a|
|No||152 (50)||133 (44)||18 (6)|
|Yes||0||19 (79)||5 (21)||< 0.0001a|
|No||164 (48)||157 (46)||24 (7)|
|Yes||22 (26)||52 (61)||12 (14)||< 0.0001a|
|No||142 (50)||124 (44)||17 (6)|
The six variables found to be significant in the univariate analysis were entered into a stepwise logistic regression modeling procedure. These included a change in size, bleeding, itch, skin breakdown, pain, and lump. The final results of the stepwise logistic regression are provided in Table 4 and demonstrate a significant association between the increasing category of Breslow depth and the presence of all of the S/Ss, except skin breakdown. The test for the proportional odds assumption was not significant (P = 0.18), providing evidence that the cumulative logit model is a reasonable model for these data. The strongest association was found between increasing depth and the presence of bleeding. For a patient who reports bleeding, the odds of having a thick versus intermediate melanoma are 7.5. Similarly, the odds of having an intermediate versus a thin melanoma for a patient reporting bleeding are 7.5. When the data were analyzed by the number of S/Ss reported, the presence of one or two of the five S/Ss or the presence of three or more S/Ss was significantly predictive of a deeper category of lesion (Table 5).
|Sign/symptom||P value||ORa||95% CI|
|Change in size||0.02||1.7||1.1–2.6|
|No. of signs/symptoms||No. of patients||Breslow depth (mm)||OR (95% CI)|
|≤ 1 (%)||> 1 and < 4 (%)||≥ 4 (%)|
|0||114||83 (73)||29 (25)||2 (2)||Referent category|
|1 or 2||193||75 (39)||107 (56)||11 (6)||4.2 (2.5–6.9)|
|3 or more||62||6 (10)||40 (65)||16 (26)||24.1 (11.5–50.3)|
Few studies have addressed the clinical implications of the S/Ss experienced by patients with cutaneous melanoma. A 1980 study by Wick et al.16 described the prevalence of S/Ss in 786 patients with melanoma. They reported that an increase in size and a change in color were the two most common S/Ss reported. Bleeding and ulceration were associated with lesions of more advanced Clark level (Breslow depth was not used at that time). We used Breslow depth instead of Clark level as the most important determinant of survival. Bleeding was associated with a higher category of Breslow depth, although skin breakdown was not. Blum et al.17 studied factors associated with a delay in treatment in 429 patients with cutaneous melanoma. They found that initial misdiagnosis of the lesion by a physician, self-detection, and head-and-neck site were associated with a delay in treatment. They also found that the three most common S/Ss of melanoma were a change in color, an increase in size, and the elevation of a pigmented lesion. The patients in that study reported a lower frequency of each symptom, except change in color, than did patients in the current study or in the Wick et al. study.16 This may be because patients were interviewed 1–3 years after diagnosis.
Other investigators have found that patients with thick melanomas are more likely to present with skin breakdown16, 18 or bleeding.16–18 Cassileth et al.18 discussed ways in which S/Ss act as catalysts that cause a patient to seek medical attention. Their study demonstrated that bleeding led patients to seek prompt medical attention and, as we have observed, was associated with the deepest lesions. Patients with symptoms of skin breakdown, itching, and pain were less concerned and commonly waited months before seeing a physician. Nonetheless, these patients were more likely to present with melanomas of intermediate thickness.
We have demonstrated that most patients with melanoma have S/Ss related to their disease. Awareness of these early S/Ss is critical for early detection of melanoma by the physician. In the current study, we found a strong association between the presence and number of S/Ss and the Breslow depth category of the primary melanoma. We believe that this finding will prompt the treating physician to take a careful history of the patient with invasive primary melanoma. By doing so and by understanding the correlation between S/Ss and Breslow depth, physicians will have a low threshold for obtaining a second pathologic opinion when the clinical history is inconsistent with the pathologist's report. In addition, in the unusual event that the Breslow depth is unavailable due to poor specimen handling or inappropriate biopsy techniques, physicians should use the additional clinical information obtained to guide appropriate therapy.
The authors thank Ms. Melissa Glim for editorial assistance.
- 15Regression models for ordinal data. J R Stat Soc Ser B. 1980; 42: 109–142..