383 Survival of Patients with Melanoma of the Lower Extremity Decreases with Distance from the Trunk Eddy C. Hsueh, M.D. Anthony Lucci, M.D. Karen Qi, M.S. Donald L. Morton, M.D. John Wayne Cancer Institute at Saint John’s Health Center, Santa Monica, California. BACKGROUND. Early stage melanoma of the lower extremity is generally associated with a favorable prognosis. However, several retrospective studies have suggested that melanoma on the foot portends poor survival. The authors hypothesized that the region of the lower extremity has prognostic importance. METHODS. Between January 1, 1971, and December 31, 1991, 652 patients were seen at the John Wayne Cancer Institute for a primary melanoma on the foot (92 patients), calf (336 patients), or thigh (224 patients). All patients had clinically or histopathologically negative regional lymph nodes. The duration of follow-up after first diagnosis was 9 –302 months, with a minimum of 6 years for survivors. Survival curves were estimated by the Kaplan–Meier method. Pearson chi-square test was used to test differences associated with the regional site of the lower-extremity melanoma. The log rank test was used for univariate analysis, and Cox proportional hazards regression was used for multivariate analysis. RESULTS. Univariate analysis identified regional site, gender, Breslow depth, Clark level, and age at diagnosis as significant for both overall survival (OS) and disease free survival (DFS) (P ⫽ 0.0001). Multivariate analysis confirmed regional site as an independent prognostic variable for OS (P ⫽ 0.0002) and DFS (P ⫽ 0.0005). Ten-year rates of OS and DFS were 71% and 66%, respectively, for patients with foot melanomas, compared with 92% and 87% for those with calf melanomas and 95% and 94% for those with thigh melanomas. CONCLUSIONS. The prognosis for patients with primary melanoma of the lower extremity is affected by the distance of the lesion from the trunk. Thus, distal (foot) lesions carry a higher risk than thigh lesions. This difference should be considered as a covariate when stratifying patients in clinical trials. Cancer 1999;85:383– 8. © 1999 American Cancer Society. KEYWORDS: melanoma, foot, extremity, survival, prognosis. Presented at the Society of Surgical Oncology/First World Congress of Surgical Oncology, San Diego, California, March 25–29, 1998. Supported in part by grants (CA12582 and CA29605, awarded to Dr. Morton) from the National Cancer Institute and by funding from the Wrather Family Foundation, Los Angeles, California. Address for reprints: Donald L. Morton, M.D., John Wayne Cancer Institute, 2200 Santa Monica Boulevard, Santa Monica, CA 90404. Received April 22, 1998; revision received July 16, 1998; accepted August 7, 1998. © 1999 American Cancer Society T he anatomic location of a primary malignant melanoma has prognostic significance. Primary lesions of the lower extremity are generally associated with a more favorable prognosis than truncal or head and neck melanomas.1,2 However, several previous studies of patients with clinically localized melanomas on the lower extremity showed that location on the foot was an independent risk factor for recurrent disease.3,4 We hypothesized that there was a correlation between survival and the regional site of a lower-extremity primary melanoma. PATIENTS AND METHODS Patient Population The study population was drawn from all patients seen at our institution between January 1, 1971, and December 31, 1991, for a lower-extremity primary melanoma corresponding to American Joint 384 CANCER January 15, 1999 / Volume 85 / Number 2 TABLE 1 Patient Demographics TABLE 2 Distribution of ELND in Three Subgroups According to Location, Breslow Depth, and Invasion of Primary Melanoma Primary site ELND Thigh (%) Calf (%) Foot (%) 224 336 92 P valuea Location Total Tumor status of regional lymph nodes Histologically negativeb Clinically negative Gender Male Female Clark levelc ⱕIII ⬎III Unknown Breslow depthc Median ⱕ1.5 mm ⬎1.5 mm Unknown Age (yrs) at diagnosis Median ⬍60 ⱖ60 Thigh 0.334 104 (46%) 120 (54%) 154 (46%) 182 (54%) 50 (54%) 42 (46%) 62 (23%) 162 (77%) 72 (21%) 264 (79%) 37 (40%) 55 (60%) 157 (70%) 60 (27%) 7 (3%) 227 (67%) 103 (31%) 6 (2%) 41 (45%) 47 (51%) 4 (4%) 0.96 149 (66%) 42 (19%) 33 (15%) 1.00 205 (61%) 94 (28%) 37 (11%) 1.40 44 (48%) 32 (35%) 16 (17%) 41 199 (89%) 25 (11%) 45 258 (77%) 78 (23%) 51 65 (71%) 27 (29%) Calf 0.001 Foot ⬍0.0001 Breslow depth Yes (%) No (%) ⱕ1.5 mm ⬎1.5 mm Unknown ⱕ1.5 mm ⬎1.5 mm Unknown ⱕ1.5 mm ⬎1.5 mm Unknown 56 (25%) 33 (15%) 15 (7%) 74 (22%) 63 (19%) 17 (5%) 18 (20%) 24 (26%) 8 (9%) 93 (42%) 9 (4%) 18 (8%) 131 (39%) 31 (9%) 20 (6%) 26 (28%) 8 (9%) 8 (9%) 53 (24%) 47 (21%) 4 (2%) 79 (24%) 72 (21%) 3 (1%) 18 (20%) 32 (35%) 0 (0%) 104 (46%) 13 (9%) 3 (1%) 148 (44%) 31 (9%) 3 (1%) 23 (25%) 15 (16%) 4 (4%) Clark level 0.003 Thigh Calf ⬍0.0001 a P value by Pearson chi-square test. “Histologically negative” refers to patients who underwent elective lymph node dissection (ELND). Patients with a tumor positive ELND specimen were excluded from the study. c Clark level was recorded for 635 patients, and Breslow depth was recorded for 566 patients. Foot ⱕIII ⬎III Unknown ⱕIII ⬎III Unknown ⱕIII ⬎III Unknown ELND: elective lymph node dissection. b Committee on Cancer Stage I or II.5 All patients were diagnosed at our institution or referred to us within 3 months of diagnosis. We did not include patients with subungual lesions because their prognosis is considered poor.6 – 8 We also excluded any patient undergoing elective lymph node dissection (ELND) if histopathologic analysis of the lymph node specimen revealed metastases. Informed consent was obtained from all patients prior to any surgical procedure. The study patients were divided into foot, calf, and thigh groups according to the regional site of the lower-extremity melanoma. After exclusion of 54 patients who had tumor positive ELND specimens (18 patients in the foot group, 25 in the calf group, and 11 in the thigh group), there were 652 study patients: 92 with primary melanoma on the foot (52 dorsal and 40 plantar), 336 with primary melanoma on the calf, and 224 with primary melanoma on the thigh. Approximately half of the patients in each group had undergone ELND (Table 1). Those who did not usually had thin (ⱕ1.5 mm Breslow depth and/or ⱕ Clark level III) primary lesions (Table 2). The study group contained a disproportionate number of females, consistent with previous reports of a higher incidence of lower-ex- tremity melanoma in women.9,10 The Breslow depth of the primary and the presenting age of the patient tended to increase as the extremity site of the primary moved from thigh to calf to foot. Seventeen patients (3%) received adjuvant polyvalent melanoma cell vaccine (5 in the foot group, 7 in the calf group, and 5 in the thigh group). The median follow-up time was 142 months (range, 9 –302 months); survivors were followed for at least 72 months. Statistical Analysis Survival curves were estimated by the nonparametric Kaplan–Meier method. Log rank test was used for univariate analysis to determine differences between curves. Overall survival (OS) was defined as the period between the diagnosis of primary melanoma and death. Disease free survival (DFS) was defined as the period between excision of a primary melanoma and pathologic or radiographic documentation of recurrence. A probability value of ⬍ 0.05 was considered significant. Multivariate analysis was performed by Cox proportional hazard regression using only the variables that were significant in univariate analysis. When applicable, Pearson chi-square test was used to test the correlation between subgroups of two vari- Survival and Lower-Extremity Melanoma/Hsueh et al. 385 TABLE 3 Correlation of Clinical Variables with Overall Survival Variables Primary site Foot Calf Thigh Gender Male Female Breslow depth ⱕ1.5 mm ⬎1.5 mm Clark level ⱕIII ⬎III Age (yrs) at diagnosis ⬍60 ⱖ60 ELND Yes No n 5-yr rate of OS 10-yr rate of OS 92 336 224 77% 94% 97% 71% 92% 95% 171 481 88% 95% 81% 93% 398 168 98% 83% 97% 76% 425 210 97% 85% 95% 81% 522 130 94% 87% 92% 82% 308 344 92% 94% 88% 92% P value (univariate) P value (multivariate) 0.0001 0.0002 0.0001 0.0093 0.0001 0.0001a 0.0001 0.1103a 0.0002 0.0359 0.1456 — OS: overall survival; ELND: elective lymph node dissection. a Regression analysis included only those patients for whom both Breslow and Clark measurements were available. ables. All statistical analyses were two-tailed and performed using Statistical Analysis System software (SAS Institute, Inc., Cary, NC). RESULTS Overall Survival Univariate analysis with the log rank test showed that OS of patients with lower-extremity melanoma was significantly (P ⫽ 0.0001) affected by the primary’s regional site, Breslow depth and Clark level, and the patient’s gender and age at diagnosis (Table 3). There was no significant difference in OS whether or not ELND was performed (P ⫽ 0.1456). There was also no significant correlation between acral lentiginous histology and OS (P ⫽ 0.737). Male patients older than 60 years with a thick primary melanoma on the foot tended to have the worst prognosis. The 5-year and 10-year OS rates for patients with a foot primary were 77% and 71%, respectively, in sharp contrast to 94% and 92% for calf primaries and 97% and 95% for thigh primaries (Fig. 1). Although there was no significant difference in OS between calf and thigh groups (P ⫽ 0.1168), there was a statistically significant difference between patients with primaries on the foot versus more proximal regions (Fig. 1). For patients with foot melanoma, there was no difference in OS for dorsal versus plantar location of the primary (P ⫽ 0.9588). Multivariate analysis confirmed regional site of a FIGURE 1. Overall survival curves, estimated by the Kaplan–Meier method for patients with melanoma of the thigh, calf, and foot, are shown. lower-extremity melanoma as an independent prognostic variable for OS (P ⫽ 0.0002) (Table 3). Also significant on multivariate analysis were gender (P ⫽ 0.0093), Breslow depth (P ⫽ 0.0001), and age at diagnosis (P ⫽ 0.0359). Figure 2 illustrates the survival differences associated with the three regional sites according to thickness of the primary lesion. It is noteworthy that pa- 386 CANCER January 15, 1999 / Volume 85 / Number 2 thickness lesions again showed shorter survival for patients with melanoma on the foot. DFS of patients with thin (ⱕ1.5 mm) lesions on the foot was similar to DFS of patients with thicker primaries on the thigh (Fig. 4). Patients with thick (⬎1.5 mm) lesions on the foot had the shortest DFS. DISCUSSION FIGURE 2. Overall survival curves are shown for patients with melanoma of the thigh, calf, and foot, with further stratification by thickness. tients with thin (ⱕ1.5 mm) lesions on the foot had a survival similar to that of patients with thicker (⬎1.5 mm) lesions on the thigh. Patients with thick (⬎1.5 mm) lesions on the foot had the worst OS. Disease Free Survival The patient’s age and gender, ELND, and the lowerextremity melanoma’s regional site, Breslow depth, and Clark level were significant with respect to DFS in univariate analysis (Table 4). There was no significant correlation between DFS and acral lentiginous histology (P ⫽ 0.6993). For patients with a foot primary, there was also no significant difference in DFS for dorsal versus plantar location (P ⫽ 0.9512). Again, the 5-year and 10-year rates were lower among patients with foot primaries (73% and 66%, respectively) than among patients with calf primaries (91% and 87%) or thigh primaries (96% and 94%) (Fig. 3). Intergroup differences in DFS were statistically significant for patients with foot versus calf primaries (P ⫽ 0.0001) and for those with calf versus thigh primaries (P ⫽ 0.0144). Of the patients who experienced recurrence, 42% (13 patients) in the foot group, 44% (17 patients) in the calf group, and 56% (9 patients) in the thigh group had regional lymph node recurrence with or without concurrent distant metastases. Multivariate analysis established foot primary as an independent prognostic variable influencing DFS (P ⫽ 0.0005) (Table 4). Breslow depth (P ⫽ 0.0001), Clark level (P ⫽ 0.0129), and age at diagnosis (P ⫽ 0.0016) were also significantly correlated with DFS on multivariate analysis. ELND and gender were not significantly correlated with DFS on regression analysis. Intergroup comparisons of patients with same- Although several studies have reported a high risk of recurrence for melanoma on the foot,3,4,11,12 their study populations were small and/or included patients with subungual melanoma, which is prognostically distinct from other extremity lesions.6 – 8 Das Gupta and Brasfield6 reported a 5-year survival rate of 38% for subungual melanoma, with a median survival of 10 months when lymph node metastases developed. Pack and Oropeza7 reported 5-year OS and DFS rates of 48% and 30%, respectively, in their study of subungual melanoma patients. Heaton et al.8 noted a 5-year OS rate of 59% for 46 subungual melanoma patients. Our study demonstrated site-dependent differences in survival for patients with lower-extremity melanomas other than subungual melanoma. There was a significant difference in OS for patients with foot versus calf or thigh primaries, and a significant difference in DFS for all three subgroups. We cannot explain the worsening of prognosis with distance from the trunk. Although the incidence of recurrence was higher in the foot group, the relative proportions of regional and distant recurrence were similar among the three groups. The higher proportion of patients undergoing ELND for a foot melanoma (54%) versus a calf melanoma (46%) or thigh melanoma (46%) argues against underdiagnosis of clinically occult regional lymph node metastases as an explanation for the poor outcome in the foot group. In addition, there was no significant correlation between ELND and survival. There is a reportedly higher incidence of acral lentiginous histology for foot melanomas,13 however several studies have shown that histologic type does not predict survival.4,14 In our study, acral lentiginous histology was not significantly correlated with either OS or DFS. Furthermore, among patients in the foot group there was no difference in survival associated with acral (plantar) versus nonacral (dorsal) melanomas. In our study population, Breslow depth tended to increase with distance from the trunk (Table 1), but comparison of patients with lesions of the same thickness showed that survival was still shorter among those with more distal lesions. Furthermore, primary location on the foot was independent of thickness in multivariate analysis. Similarly, although lower-extremity primary site was associated with gender and age (Table 1), it was also an independent prognostic variable in multivariate analysis. Survival and Lower-Extremity Melanoma/Hsueh et al. 387 TABLE 4 Correlation of Clinical Variables with Disease Free Survival Variables Primary site Foot Calf Thigh Gender Male Female Breslow depth ⱕ1.5 mm ⬎1.5 mm Clark level ⱕIII ⬎III Age (yrs) at diagnosis ⬍60 ⱖ60 ELND Yes No n 5-yr rate of DFS 10-yr rate of DFS 92 336 224 73% 91% 96% 66% 87% 94% 171 481 84% 92% 79% 89% 398 168 96% 76% 95% 68% 425 210 96% 78% 94% 72% 522 130 92% 80% 89% 74% 308 344 87% 92% 83% 90% P value (univariate) P value (multivariate) 0.0001 0.0005 0.0003 0.0652 0.0001 0.0001a 0.0001 0.0129a 0.0001 0.0016 0.0254 0.3459 DFS: disease free survival; ELND: elective lymph node dissection. a Regression analysis included only those patients for whom both Breslow and Clark measurements were available. FIGURE 4. Disease free survival curves are shown for patients with melaFIGURE 3. noma of the thigh, calf, and foot, with further stratification by thickness. There is no adequate anatomic or physiologic explanation for poorer survival of patients with primary melanomas on the foot versus other lower-extremity sites. Delay in diagnosis has been noted for patients with acrally located melanoma.15 This may explain the thicker melanomas found in the foot group than in the other two groups (Table 1). The longer duration of active disease may contribute to the worse survival in this group. Variations in rates of lymphatic flow sec- ondary to repeated tissue compression in a weightbearing area have been observed.16 Compression may enhance the lymphatic passage of tumor cells from a primary melanoma on the foot, thereby increasing the incidence of disease recurrence. However, there is little evidence to implicate this as the cause of worse prognosis for the foot group. In summary, lower-extremity melanoma can be prognostically stratified on the basis of regional site (foot versus calf versus thigh), with foot lesions carry- Disease free survival curves, estimated by the Kaplan–Meier method for patients with melanoma of the thigh, calf, and foot, are shown. 388 CANCER January 15, 1999 / Volume 85 / Number 2 ing the highest risk of recurrence and the lowest rates of survival. Meticulous cutaneous examination of the foot, including the interdigital web space, is vital for early detection of foot melanoma. Because of the significant difference in survival between foot melanoma and other lower-extremity melanoma (exclusive of subungual melanoma), future adjuvant trials involving lower-extremity melanoma should consider site of primary in the leg as a stratification factor. 6. 7. 8. 9. 10. REFERENCES 1. 2. 3. 4. 5. Lewis MH, Hill JT, Leopold JG, Hughes LE. Guidelines for management of malignant melanoma of the lower limb based on a study of long-term behaviour. Clin Oncol 1982; 8:341–9. Balch CM, Soong S-J, Shaw HM, Urist MM, McCarthy WH. An analysis of prognostic factors in 8500 patients with cutaneous melanoma. In: Balch CM, Houghton AN, Milton, GW, Sober AJ, Soong S-J, editors. Cutaneous melanoma. 2nd edition. Philadelphia: J. B. Lippincott, 1992:165. Day CL Jr., Sober AJ, Kopf AW, Lew RA, Mihm MC Jr., Hennessey P, et al. A prognostic model for clinical stage I melanoma of the lower extremity: location on foot as independent risk factor for recurrent disease. Surgery 1981;89: 599 – 603. Urist MM, Balch CM, Soong S, Shaw HM, Milton GW, Maddox WA. The influence of surgical margins and prognostic factors predicting the risk of local recurrence in 3445 patients with primary cutaneous melanoma. Cancer 1985;55: 1398 – 402. Fleming ID, Cooper JS, Henson DE, Hutter RVP, Kennedy BJ, Murphy GP, et al., editors. AJCC cancer staging manual. 5th edition. Philadelphia: Lippincott-Raven, 1997. 11. 12. 13. 14. 15. 16. Das Gupta T, Brasfield R. Subungual melanoma: 25-year review of cases. Ann Surg 1965;161:545–52. Pack GT, Oropeza R. Subungual melanoma. Surg Gynecol Obstet 1967;124:571– 82. Heaton KM, el-Naggar A, Ensign LG, Ross MI, Balch CM. Surgical management and prognostic factors in patients with subungual melanoma. Ann Surg 1994;219:197–204. Fortin PT, Freiberg AA, Rees R, Sondak VK, Johnson TM. Malignant melanoma of the foot and ankle. J Bone Joint Surg Am 1995;77:1396 – 403. Hutchinson BL. Malignant melanoma in the lower extremity: a comprehensive overview. Clin Podiatr Med Surg 1986; 3:533–50. Day CL Jr., Lew RA, Mihm MC Jr., Sober AJ, Harris MN, Kopf AW, et al. A multivariate analysis of prognostic factors for melanoma patients with lesions greater than or equal to 3.65 mm in thickness. The importance of revealing alternating Cox models. Ann Surg 1982;195:44 –9. Barnes BC, Seigler HF, Saxby TS, Kocher MS, Harrelson JM. Melanoma of the foot. J Bone Joint Surg Am 1994;76:892– 8. Feibleman CE, Stoll H, Maize JC. Melanomas of the palm, sole, and nailbed: a clinicopathologic study. Cancer 1980;46: 2492–504. Sondergaard K, Olsen G. Malignant melanoma of the foot: a clinicopathological study of 125 primary cutaneous malignant melanomas. Acta Pathol Microbiol Scand (A) 1980;88: 275– 83. Metzger S, Ellwanger U, Stroebel W, Schiebel U, Rassner G, Fierlbeck G. Extent and consequences of physician delay in the diagnosis of acral melanoma. Melanoma Res 1998;8: 181– 6. Ikomi F, Hunt J, Hanna G, Schmid-Schonbein GW. Interstitial fluid, plasma protein, colloid, and leukocyte uptake into initial lymphatics. J Appl Physiol 1996;81:2060 –7.