A multifactorial analysis of 200 cutaneous melanoma patients with distant metastasis (stage III) was performed on 13 clinical and pathological factors using the Cox regression analysis. There were only three dominant prognostic variables that independently predicted the patient's clinical course: (1) number of metastatic sites (1 vs. 2 vs. greater than or equal to 3, p less than 0.00001), (2) remission duration (less than 12 mo vs. greater than or equal to 12 mo, p = 0.0186), and (3) the location of the metastases (visceral vs. nonvisceral vs. combined, p = 0.0192). Factors that were not significant in the multifactorial analysis included the patients' age and sex, the site of the primary melanoma, the sequence of metastases, and all histopathological features of the primary melanoma (thickness, level of invasion, ulceration, growth pattern, pigmentation, and lymphocyte infiltration). For a single metastatic site, the 1-yr survival rate was 36%, while it was only 13% for 2 sites, and 0% for greater than or equal to 3 sites (p less than 0.00001). The 1-yr survival for patients was 40% for nonvisceral sites (skin, subcutaneous, distant lymph nodes) compared to only 11% for visceral metastases and 8% for combined sites (p less than 0.00001). Pulmonary metastases were associated with a significantly higher survival rate than metastatic melanoma in any other visceral site. The most common first site of distant metastases (either alone or in combination) was skin (38%), lung (36%), liver (20%), and brain (20%). The skin, subcutaneous and distant lymph node group was the first site of metastases in 59% of patients. This finding emphasizes the importance of careful physical exams in routine metastatic evaluations. Only a minority (25%) of stage I patients progressed to stage III disease after a median interval of 2.8 years. In contrast, the majority (75%) of melanoma patients with nodal metastases (stage II) progressed to stage III disease after a median duration of only 11 mo. Of the patients who eventually developed stage III disease, 95% of those who initially presented with stage II disease progressed within 3 yr, while stage I patients who progressed to stage III did not reach a 95% cumulative incidence until 8 yr.

Ulceration of a cutaneous melanoma on microscopic sections is an adverse prognostic finding. The five-year survival rate is reduced from 80% for non-ulcerated melanomas to 55% in the presence of ulceration for Stage I melanoma patients and from 53 to 12% for Stage II melanoma patients (P less than 0.001). As a group, ulcerated lesions are thicker and more likely to have a nodular growth pattern. However, survival rates were still worse for ulcerated melanomas when matched with nonulcerated lesions for thickness and stage of disease. The width but not the depth of surface ulceration significantly correlated with survival. The median ulcer depth was 0.08 mm (range 0.01-1.2 mm). In those few lesions with ulcer craters more than 0.2 mm in depth, the melanomas were so thick they had the same poor prognosis regardless of whether thickness was measured to the base of the ulcer or to the top of the lesion. The Breslow microstaging method of measuring thickness is therefore a valid prognostic indicator, even for ulcerated lesions. The incidence of ulceration for the entire patient group ranged from 12.5% for melanomas less than 0.76 mm thickness to 72.5% for melanomas greater than 4.0 mm thick (P of correlation = 0.0001); from 12% for Level II invasion to 63% for Level V lesions (P = 0.005); from 23% for superficial spreading growth patterns to 49% for nodular and 74% for polypoid lesions (P = 0.0001); and from 27% for lesions with a heavy lymphocyte infiltration to 60% for minimal or absent host response (P = 0.005). There was no significant correlation with anatomic location, pigmentation of the melanomas, or with the patient's age and sex. Since ulceration appears to have such an important influence on survival rates, this parameter should be considered as a stratification criterion in clinical trials and accounted for when analyzing results of melanoma treatment.

Risk factors associated with local recurrences were analyzed from a series of 3445 clinical Stage I melanoma patients. In single‐factor analysis, tumor thickness, ulceration, and increasing age were highly significantly predictive of recurrence (p < 0.00001). After 5 years of follow‐up, local recurrence rates were 0.2% for tumors less than 0.76 mm thick, 2.1% for tumors 0.76 to 1.49 mm thick, 6.4% for tumors 1.5 to 3.99 mm thick, and 13.2% for tumors 4.0 mm or greater in thickness. Ulcerated melanomas recurred more often than nonulcerated lesions (11.5% versus 1.9%). When analyzed as a continuous variable, increasing age increased the risk of local failure. In multifactorial analysis, all of these three factors remained independently predictive of local recurrence. Recurrences were more common with nodular melanomas (5.6%) compared to superficial spreading (2.5%) or lentigo maligna melanoma (2.5%), but this difference did not reach statistical significance (P = 0.115). Lower extremity (4.7%) and head and neck lesions (4.4%) recurred more frequently than upper extremity (1.6%) or trunk (1.2%) melanomas (P = 0.0217). The highest recurrence rates were observed in patients with melanomas located on the foot (11.6%) and hand (11.1%). The safety of conservative margins for the excision of low‐risk melanomas was demonstrated in a review of 1151 consecutive patients with melanomas less than 1 mm thick where only one local recurrence was observed. Sixty‐two percent of these patients had resection margins of 2 cm or less. In 95 patients local recurrence developed as the first site of relapse and were treated with surgical excision. The median survival for this group was 3 years, whereas 20% of this group survived 10 years. These data demonstrate that: (1) the risk of local recurrence rises with increasing tumor thickness, presence of ulceration, and age; (2) melanomas less than 1 mm thick have a very low local recurrence rate, even when excised with margins of 2 cm or less; and (3) local recurrence is a poor prognostic sign because regional and systemic metastases subsequently develop in many patients. Cancer 55:1398‐1402, 1985.

Twelve clinical and pathologic parameters were compared in two series of Stage I melanoma patients treated at the University of Alabama in Birmingham, USA (676 patients) and at the University of Sydney in New South Wales, Australia (1,110 patients). Actuarial survival rates were virtually the same at the two institutions over a 25-year follow-up period. The incidence of thin melanomas (less than 0.76 mm) was also similar at both geographic locations (25% vs. 26%). Other similarities of these two patient populations included the following: 1) tumor thickness (Breslow Microstaging). 2) level of invasion (Clark Microstaging), 3) surgical results, 4) sex distribution, and 5) age distribution. The greatest differences between the two patient populations were their 1) anatomic distribution, 2) growth pattern, and 3) incidence of ulceration. The trunk was the most common site of melanoma, and occurred more frequently among Australian patients (37% vs. 28%). A multifactorial analysis (Cox's regression model) was then performed that included a comparison of the two institutions as a variable (Alabama vs. Australia). The dominant prognostic factors (p less than 0.0001) were 1) ulceration, 2) tumor thickness, 3) initial surgical management (wide excision +/- node dissection), 4) anatomic location, 5) pathologic stage (I vs. II), and 6) level of invasion. The benefit of elective lymph node dissection was demonstrated in both series for patients with intermediate thickness melanoma (0.76 to 3.99 mm.) For melanomas ranging from 0.76 to 1.5 mm in thickness, the benefit of node dissection was primarily in male patients. Survival rates for melanoma at the two institutions were not significantly different in the multifactorial analysis, even after adjusting for all other variable. Thus, the biologic behavior of melanoma in these two different parts of the world was virtually the same, with only minor differences that did not significantly influence survival rates. Long-term follow-up exceeding eight to ten years after surgery is critical in the interpretation of these prognostic factors and the surgical results.