- Email: [email protected]
Prognostic factors in patients with regional cervical nodal metastases from cutaneous malignant melanoma
Prognostic Factors in Patients With Regional Cervical Nodal Metastases From Cutaneous Malignant Melanoma
S. Eva SingMary, MD, R&etl M. Byers, MD, Richard Shaknbeqpr, Guinee, MD, MPH, Houston, Texas
Histologic evidence of involvement of regional lymph nodes by malignant melanoma from a primary tumor in the head and neck region carries an ominous prognosis. Even though local control of the disease is often achieved with regional node dissection, the patient usually dies from systemic hematogenous metastases. Identification of prognostic features within this group of patients may allow more selective staging to predict clinical outcome and aid in the accurate assessment of the value of adjuvant modalities. To determine the existence of any significant predictive variables of the biologic behavior of melanoma after the development of regional disease, we conducted a detailed retrospective analysis of the clinical records of patients treated at the University of Texas-M.D. Anderson Hospital and Tumor Institute at Houston between 1944 and 1974. All patients who had development of histologically positive regional cervical nodal metastases from cutaneous melanoma during the course of their disease were included, with a minimum 10 year follow-up. Material and Methods Between March 1944 and December 1974, 4,322 patients with the diagnosis of malignant melanoma were treated at our institution. Of these patients, 1,169 had development of regional nodal metastases during the course of their disease, of whom 324 had cervical nodal metastases. In this latter group, radical or modified neck dissections were performed on 301 patients, cervical node biopsies were performed on 21 patients, and surgical treatment was refused by 2 patients. This retrospective analysis was composed of the 287 patients who underwent From the Departments of Sugery. Head andNeck Surgery, and Patient Studies. University of Texas4.D. Anderson Hospital and Tumor InstiMe at Houston. Houston, Texas. Requests for reprints should be addressed to S. Eva Singletary, MD. Department of Surgery, Universtty of Texas4.D. Anderson Hospital and Tumor Institute. PD Box 106,6723 Bertner Avenue, Houston, Texas 77030. Presented at tb 32nd Annual Meeting of the Society of Head and Neck Surgeons, Colorado Springs. Colorado, May 7-10. 1966.
Volume 152, October 1966
MPH, Charles M. McBride, MD, and Vhcent F.
neck dissection for histologically positive nodal metastases from cutaneous melanoma, excluding those who had
node biopsies, those who refused treatment, and those with primary mucosal melanoma. Cumulative survival rates by the life table method were calculated from the date of nodal dissection to the date of last follow-up or death. The p values were determined by Lee-Desu statistical analysis. Results
In this series, 202 patients (70 percent) were male and 85 (30 percent) were female. The age and sex distribution of these patients is shown in Figure 1. The majority of patients were in their sixth decade, but they ranged in age from 15 to 89 years. All patients were white. Sites of the primary tumors included the scalp in 65 patients (23 percent), an ear in 21 patients (7 percent), the face in 81 patients (28 percent), the neck in 51 patients (18 percent), the upper trunk in 23 patients (8 percent), unknown sites of primary tumor in 45 patients (16 percent), and multiple primary sites in 1 patient. Men had a higher incidence of primary tumors of the scalp (26 percent versus 15 percent in women) and neck (27 percent versus 15 percent in women). Women had a slightly increased incidence of primary tumors of the face (32 percent versus 27 percent in men), upper trunk (11 percent versus 7 percent in men), and unknown sites of origin (18 percent versus 14 percent in men). Men and women had almost equal involvement of the ear (8 percent in men versus 7 percent in women). The primary tumor was absent secondary to previous treatment in 178 patients (62 percent). Only 28 patients (10 percent) presented with an untreated primary tumor. In another 29 patients (10 percent), the primary tumor was recurrent, and in 7 patients (2 percent) satellitosis was present within 3 cm of the primary tumor. The primary melanomas were treated by excision and closure in 175 patients
371
Singletary et al
70 -
TABLE I
60 E 50 = a 6 40 -
Patient Characteristics
B 30 z’
0 l-9
m-10
2*20
30-32
4*49
*so
2.042
TO-TO
m-32
Age (Years) Figure 1. Age and wx dktributkm of 202 msle patkwds (shaded columns) and 85 fern& psllontr (w/t/b columns) who lmdsrwontmodn&d or radlcai & dhsecthw for reghmal nods/ metastaae8 of melanoma.
(61 percent), excision and skin graft in 43 (15 percent), cauterization in 14 (5 percent), and irradiation in 10 (3 percent). In 45 patients ( 16 percent), no physical findings or clinical history could document the existence of a primary tumor. Only nine patients (3 percent) received adjuvant treatment of the primary tumor, which consisted of chemotherapy in two patienta, immunotherapy in four patienta, irradiation in one patient, and a combination of chemotherapy and immunotherapy in two patients. Because the majority of the melanomas were treated before referral to our institution, the clinical characteristics of the primary tumor were not adequately documented for statistical analysis. Also, microstaging by Clark’s levels of invasion or Breslow’s thickness of the primary tumor was not carried out during the time period of this review. Patients were categorized according to nodal treatment (Table I). Twenty-seven patients (9 percent) underwent elective cervical node dissections from which clinically uninvolved but histologically involved nodes were obtained. In 53 patients (19 percent), immediate therapeutic node dissection of clinically and histologically positive nodes was performed, and in 162 patients (56 percent), a delayed therapeutic node dissection was required for clinically and histologically involved nodes that developed 2 months or more after the primary tumor was treated. Node dissection was therapeutically indicated in 45 patienta (16 percent) for whom no site of the primary melanoma was detected. The sites of the primary tumor were similar for each type of node dissection (Table I). The exception was a predominance of scalp tumors in patients with immediate therapeutic node dissection (40
372
Comparison ol Patknt Characterldlce by Type of Nodal Dtssectlon (values expressed as percentages)
Total patients (n) Site of primary lesion Scalp Ear Face Neck Upper trunk Multiple sites Unknown site Clinical nodal status Uninvolved Single, <3 cm Multiple, <3 cm Fixed Bilateral Biopsied Not described Involved nodes (n) 1 only 2-4 5-10 >lO Matted Not described Extranodal invasion Adjuvant nodal therapy
Elective
Dlssection of Nodes Unknown ImmeDeprimary diate layed Site
27
53
162
45
22 11 37 30 0 0 0
40 2 23 19 15 1 0
24 ;: 20 9 0 0
0 0 0 0 0 0 100
100 0 0 0 0 0 0
0 19 30 8 2 26 15
0 28 15 8 0 36 13
0 7 9 9 0 71 4
48 37 7 4 0 4 0 11
19 26 17 23 6 9 11 13
38 32 11 11 2 6 10 17
29 36 13 9 4 9 9 17
percent), as compared with those who had elective (22 percent) and delayed therapeutic (24 percent) node dissection. Of the patients who underwent an immediate therapeutic node dissection, the clinical nodal status (Table I) was recorded as a single node of less than 3 cm in 19 percent. In contrast, patients who underwent a delayed therapeutic node dissection had a higher incidence of a single involved node (28 percent). The number of histologically involved nodes was highest in patients who had immediate therapeutic node dissection and lowest in patients who underwent elective node dissection (Table I). One involved node was found in 48 percent of the patients with elective dissection, 38 percent of thoee with delayed therapeutic dissection, 29 percent of those with therapeutic dissection in whom the primary site was unknown, and 19 percent of those with immediate therapeutic dissection. Documented histologic extranodal invasion was absent in all patients who had elective node dissections, but occurred in a similar percentage of patients with immediate therapeutic dissection (11 percent), delayed therapeutic dissection (10 percent), and therapeutic dissection for metastases from an unknown primary site (9 percent) (Table I). Regional nodal recurrence after a neck dissection
Cervical Nodal M&stases From Cutaneous Melanoma
with a histologically involved node or nodes occurred in 42 patients (15 percent). Individually, factors such as sex and age of the patient, clinical stage at time of first evaluation at our hospital, site of the primary tumor, elective versus therapeutic node dissection, and type of adjuvant treatment after surgery did not influence nodal recurrence in the neck. The median age of the patients for whom nodal surgery failed was 43 years (range 17 to 83 years). Of these, 29 were male (14 percent) and 13 were female (15 percent). Regional therapy failed in 15 percent of the patients who initially presented at our hospital with cervical nodal metastases and underwent immediate neck dissection. This rate was not significantly higher than that for patients with disease confined to the primary site who had subsequent development of nodal metastases (9 percent). Of interest, but not of statistical significance, is the fact that 19 percent of the patients who had an elective node dissection exhibited recurrent disease in the neck dissection field compared with 17 percent of those who had immediate therapeutic dissection, 14 percent of those who had delayed therapeutic dissection, and 11 percent of those who had therapeutic dissection with an unknown primary site. The number of involved nodes did correlate with the rate of local therapy failure, however. Recurrent disease in the neck developed in 9 percent of the patients with only one histologically involved node. If 2 to 4 nodes were involved, 10 percent of the patients experienced a recurrence; however, if 5 to 10 nodes were involved or if more than 10 involved nodes were found, the recurrence rate increased to 22 percent and 26 percent, respectively. The highest recurrence rate (44 percent) was seen in patients whose involved nodes were described as matted. After neck dissection, in-transit disease occurred in 12 patients, consisting of primary tumors of the face in 7 patients, of the scalp in 3, of the upper trunk in 1, and of the neck in 1. Elective node dissections had been performed in four of these patients and delayed therapeutic dissection in eight. Nine of these patients had multiple involved nodes, and in seven, concomitant recurrent cervical nodal disease developed. Only 75 patients in this series (26 percent) remained free of disease after neck dissection. In 28 patients (10 percent), disease status at the time of death was unknown. The remaining 164 patients (64 percent) had documented distant metastases. Distant metastases were most common in patients with primary tumors of the scalp (77 percent), an ear (71 percent), the neck (63 percent), and the upper trunk (78 percent), and lowest in patients with primary tumors of the face (54 percent) or an unknown site (45 percent). There was no significant difference in the frequency of distant me&stases based on the timing of neck dissection in relation to treatment of
Volume 152, Octabm 1996
0
2
4
6
8
10
YeUS Fme 2. Swvlval rates from @e t&no of mMa dhsecthn to last foffow-up or desth. Cum A tdcates owratt auvtvaL Curva 6 Indicates mvfval wftf~d&ant m&stases after nods i%sssctkm
the primary tumor. The occurrence of distant metastases was lowest when only one histologically involved regional node was found (56 percent) and highest when there were multiple matted regional nodes (100 percent). Distant disease developed in 81 percent of patients with extranodal disease at the time of neck dissection. The firat sites of distant metastases were, in order of frequency, simultaneous multiple sites in 50 patients (27 percent), a lung in 41 (22 percent), the brain in 27 (15 percent), subcutaneous tissue in 23 (12 percent), the liver in 13 (7 percent), bone in 7 (4 percent), distant lymph nodes in 3 (2 percent), and the gastrointestinal tract in 3 (2 percent). In 17 patients (9 percent), the site of metastasis was known but not stated. The median length of survival from diagnosis of initial distantmetastases to the date of last follow-up or death was 21 months for the patients with distant lymph node involvement, 10 months for those with bone involvement, 5 months for those with lung and subcutaneous tissue involvement, 3 months for those with gastrointestinal tract involvement, and 2 months for those with brain or liver involvement or simultaneous multiple sites of metastases or nonspecified metastases. Only 2 of these 184 patients with distant metastasee remained free of disease after treatment (1 patient who underwent surgical resection of a solitary lung metastasis and 1 patient who received combination chemotherapy and immunotherapy for bilateral pulmonary lesions, with an 18 and 10 year follow-up, respectively). The 5 and 10 year survival rates for the entire group of 287 patients was 33 percent and 28 percent, respectively. Of the 184 patients (64 percent) who had subsequent development of documented distant me&stases after neck dissection, the 5 and 10 year survival rates decreased to 13 percent and 7 percent, respectively (Figure 2). No statistical difference in survival was detected based on any one of the following factors: age, sex, treatment of the primary tumor, clinical stage at first evaluation, or adjuvant treatment of the primary tumor or nodal
373
Singletaty el al
1.0 if
23
&cl .6 g
.4
d
A
3
n. ‘s
a .2
.O
c
0
2
4
6
a
10
Flpure 3. Survival raies fromthe the of node dbsectkm to lasf follow-up o.rdeath basedon the mmber oiInvolved nodea Curve A Indloates those paflents wtlh one Involved node: Curve 8, two
to lourInvolvednodes;CurveC, flve
or more Involved nodes; and
Curve D, matted nodes.
metastases. The length of survival of patients with a primary tumor of the head and neck region was also similar to that of patients with a primary tumor of the upper trunk (p = 0.792). The median survival from the onset of nodal metastases was 15 months for primary tumors of the scalp, 14 months for primary tumors of the ear, 19 months for those of the face, 18 months for those of the neck, and 20 months for those of the upper trunk. However, patients with primary tumors of unknown sites with cervical nodal metastases had significantly better lengths of survival (median 90 months) compared with patients with primary tumors of the head and neck (p = 0.002) or the upper trunk (p = 0.020). Survival was not affected by whether the indication for the node dissection was elective or therapeutic. Only 4 percent of the 162 patients who underwent delayed therapeutic node dissection had development of nodal metastases 24 months or more after treatment of the primary tumor. The median survival of this small number of patients was 24 months, compared with a median survival of 15 months for patients with nodal metastases that developed less than 24 months after the primary tumor was treated; however, this difference in survival was not statistically sigriificant. The numtir”bf involved nodes detected by neck dissection also influenced survival (Figure 3). Patients with only one involved node had a longer survival (median 39 months) than patients with two to four involved nodes (median 24 months, p = 0.021). A marked difference in length of survival was seen between patients with one involved node and those with five or more involved nodes (median 20 months, p = 0.0003) or matted nodes (median 8 months, p = 0.0001). No statistical difference in survival was detected between patients with two to four involved nodes versus five or more involved nodes (p = 0.099). The absence of extranodal disease significantly improved length of survival ((median 30 months, p = 0.0007) in contrast to that of
374
patients in whom melanoma was present in the connective tissue at dissection (median 11 months). The percentage of involved nodes (the number of involved nodes divided by the total number of nodes removed X 100 percent) was not predictive of survival. Less than 10 percent of the total number of cervical nodes removed were positive in 56 percent of the patients who died from disseminated melanoma. Regional failure after neck dissection also carried a grim prognosis. Of the 42 patients in whom nodal recurrence developed after a neck dissection, 26 percent survived 2 years and 7 percent survived 5 years. Of the 12 patients with in-transit disease, 31 percent survived 2 years and none survived 5 years. Thus, the three major factors that had an independent, statistically significant positive effect on the survival of patients with cervical nodal metastases were an unknown primary site, one histologically involved node, and the absence of extranodal disease.
The 33 percent cumulative 5 year survival rate of the 287 patients with cervical nodal metastases indicates that histologic evidence of involvement of nodes by malignant melanoma usually signifies disseminated disease. Even more dismal, the 13 percent 5 year survival of the 184 patients who had subsequent development of distant failure after regional node dissection testifies to the virulent behavior of melanoma. Although retrospective studies have shown inherent limitations, an understanding of the results of historical treatment may provide direction for future randomized, prospective clinical trials. Most important, the stratification of patients with nodal metastases by prognostic variables may identify certain subgroups of patients whose survival rates could be improved with regional or systemic adjuvant therapy. As confirmed by others [j-4], the dominant prognostic feature after the onset of regional nodal metastases is tumor burden, that is, the number of nodes affected by metastaaes. In this review, the best survival rate was seen in patients with tumor burden limited to a single histologically involved node. If the tumor extended extranodally, the incidence of both regional and distant failure increased dramatically. Age, sex, site of known primary disease, clinical stage at presentation, elective or therapeutic node dissection, and the time interval between treatment of the primary tumor and the development of nodal metastases were not predictive of survival after nodal disease was present. However, in patients with unknown primary sites, survival was significantly higher when compared with that of patients with known primary sites. Although some studies [5-71 have reported a similar favorable outcome for patients with unknown
The Amwlcan Journal ol Surgery
Cervical Nodal Metastases From Cutaneous Melanoma
primary sites, conflicting studies have shown a less favorable outcome or no change at all [a--101. Other possible prognostic criteria not evaluated in this review were the presence or absence of ulceration and Breslow’s tumor thickness of the primary tumor. In patients with histologically positive nodes, Balch et al [1] obtained a less than 15 percent 5 year survival rate in patients with ulcerated primary cutaneous melanoma in contrast to a 30 percent 5 year survival rate in patients with nonulcerative primary tumors. Day et al [II] found Breslow’s tumor thickness of the primary tumor to be predictive of survival in patients with clinically negative nodes and metastatic involvement limited to a positive node percentage of less than 20 percent. Only by clearly identifying comparable subgroups of patients with histologically involved nodes can valid conclusions be drawn as to optimal treatment. Summary A retrospective analysis with a minimum 10 year follow-up was performed on 287 patients who underwent radical or modified neck dissections with histologically involved regional nodal metastases from cutaneous malignant melanoma. The cumulative 5 year and 10 year survival rates calculated from the time of node dissection were 33 percent and 28 percent, respectively. Age and sex of the patient, site of known primary tumor, clinical stage at presentation, elective versus therapeutic node dissection, and time interval from the treatment of the primary tumor to node dissection did not independently affect survival. However, an unknown site of primary disease, the presence of only one histologi-
volume 152, octobw lB86
cally involved node, and the absence of extranodal tumor invasion at the time of node dissection were statistically significant individual prognostic factors for an improved survival rate. References 1. Balch CM, Soong SJ, Murad TM, lngalls AL, Maddox WA. A multifactorial analysis of melanoma. Ill. Prognostic factors in melanoma patients with lymph node metsstases (stage Ill). Ann Surg 1981;193:377-88. 2. Cohen MH, Ketcham AS, Felix FI, et al. Prognostic factors in patients undergoing lymphadenectomy for malignant melanoma. Ann Surg 1977;188:835-42. 3. Fortner JG, Woodruff J, Schottenfeld D, MacLean B. Blostatistical basis of elective node dissection for malignant melanoma. Ann Surg 1977;186:101-3. 4. Karakousis CP, Seediq MK, Moore R. Prognostic value of lymph node dissection in malignant melanoma. Arch Surg 1980;115:719-22. 5. Baab GH, McBride CM. Malignant melanoma; the patient with an unknown site of primary origin. Arch Surg 1975;llO: 896-900. 6. Lopez R, Holyoke ED, Moore RH, Karakousis CP. Malignant melanoma with unknown primary site. J Surg Oncol 1982; 19:151-4. 7. Santini H, Byers RM, Wolf PF. Metastatic melanoma to cervical and parotid nodes from an unknown primary site. Am J Surg 1985;150:510-2. 8. Milton GW, Shaw HM, McCarthy WH. Occult primary malignant melanoma, factors influencina survival. Br J Surg 1977;64:805-8. 9. Das Gupta T, Bowden L, Berg J. Malignant melanoma of unknown primary origin. Surg Gynecol Obstet 1963;117: 341-50. 10. Giuliano AE, Moseley HS, Morton DC. Clinical aspects of unknown primary melanomas. Ann Surg 1980;191:98104. 11. Day CL, Sober AJ, Lew RA, et al. Malignant melanoma patients with positive nodes and relatively good prognoses. Cancer 1981;47:955-62.
375
S. Eva SingMary, MD, R&etl M. Byers, MD, Richard Shaknbeqpr, Guinee, MD, MPH, Houston, Texas
Histologic evidence of involvement of regional lymph nodes by malignant melanoma from a primary tumor in the head and neck region carries an ominous prognosis. Even though local control of the disease is often achieved with regional node dissection, the patient usually dies from systemic hematogenous metastases. Identification of prognostic features within this group of patients may allow more selective staging to predict clinical outcome and aid in the accurate assessment of the value of adjuvant modalities. To determine the existence of any significant predictive variables of the biologic behavior of melanoma after the development of regional disease, we conducted a detailed retrospective analysis of the clinical records of patients treated at the University of Texas-M.D. Anderson Hospital and Tumor Institute at Houston between 1944 and 1974. All patients who had development of histologically positive regional cervical nodal metastases from cutaneous melanoma during the course of their disease were included, with a minimum 10 year follow-up. Material and Methods Between March 1944 and December 1974, 4,322 patients with the diagnosis of malignant melanoma were treated at our institution. Of these patients, 1,169 had development of regional nodal metastases during the course of their disease, of whom 324 had cervical nodal metastases. In this latter group, radical or modified neck dissections were performed on 301 patients, cervical node biopsies were performed on 21 patients, and surgical treatment was refused by 2 patients. This retrospective analysis was composed of the 287 patients who underwent From the Departments of Sugery. Head andNeck Surgery, and Patient Studies. University of Texas4.D. Anderson Hospital and Tumor InstiMe at Houston. Houston, Texas. Requests for reprints should be addressed to S. Eva Singletary, MD. Department of Surgery, Universtty of Texas4.D. Anderson Hospital and Tumor Institute. PD Box 106,6723 Bertner Avenue, Houston, Texas 77030. Presented at tb 32nd Annual Meeting of the Society of Head and Neck Surgeons, Colorado Springs. Colorado, May 7-10. 1966.
Volume 152, October 1966
MPH, Charles M. McBride, MD, and Vhcent F.
neck dissection for histologically positive nodal metastases from cutaneous melanoma, excluding those who had
node biopsies, those who refused treatment, and those with primary mucosal melanoma. Cumulative survival rates by the life table method were calculated from the date of nodal dissection to the date of last follow-up or death. The p values were determined by Lee-Desu statistical analysis. Results
In this series, 202 patients (70 percent) were male and 85 (30 percent) were female. The age and sex distribution of these patients is shown in Figure 1. The majority of patients were in their sixth decade, but they ranged in age from 15 to 89 years. All patients were white. Sites of the primary tumors included the scalp in 65 patients (23 percent), an ear in 21 patients (7 percent), the face in 81 patients (28 percent), the neck in 51 patients (18 percent), the upper trunk in 23 patients (8 percent), unknown sites of primary tumor in 45 patients (16 percent), and multiple primary sites in 1 patient. Men had a higher incidence of primary tumors of the scalp (26 percent versus 15 percent in women) and neck (27 percent versus 15 percent in women). Women had a slightly increased incidence of primary tumors of the face (32 percent versus 27 percent in men), upper trunk (11 percent versus 7 percent in men), and unknown sites of origin (18 percent versus 14 percent in men). Men and women had almost equal involvement of the ear (8 percent in men versus 7 percent in women). The primary tumor was absent secondary to previous treatment in 178 patients (62 percent). Only 28 patients (10 percent) presented with an untreated primary tumor. In another 29 patients (10 percent), the primary tumor was recurrent, and in 7 patients (2 percent) satellitosis was present within 3 cm of the primary tumor. The primary melanomas were treated by excision and closure in 175 patients
371
Singletary et al
70 -
TABLE I
60 E 50 = a 6 40 -
Patient Characteristics
B 30 z’
0 l-9
m-10
2*20
30-32
4*49
*so
2.042
TO-TO
m-32
Age (Years) Figure 1. Age and wx dktributkm of 202 msle patkwds (shaded columns) and 85 fern& psllontr (w/t/b columns) who lmdsrwontmodn&d or radlcai & dhsecthw for reghmal nods/ metastaae8 of melanoma.
(61 percent), excision and skin graft in 43 (15 percent), cauterization in 14 (5 percent), and irradiation in 10 (3 percent). In 45 patients ( 16 percent), no physical findings or clinical history could document the existence of a primary tumor. Only nine patients (3 percent) received adjuvant treatment of the primary tumor, which consisted of chemotherapy in two patienta, immunotherapy in four patienta, irradiation in one patient, and a combination of chemotherapy and immunotherapy in two patients. Because the majority of the melanomas were treated before referral to our institution, the clinical characteristics of the primary tumor were not adequately documented for statistical analysis. Also, microstaging by Clark’s levels of invasion or Breslow’s thickness of the primary tumor was not carried out during the time period of this review. Patients were categorized according to nodal treatment (Table I). Twenty-seven patients (9 percent) underwent elective cervical node dissections from which clinically uninvolved but histologically involved nodes were obtained. In 53 patients (19 percent), immediate therapeutic node dissection of clinically and histologically positive nodes was performed, and in 162 patients (56 percent), a delayed therapeutic node dissection was required for clinically and histologically involved nodes that developed 2 months or more after the primary tumor was treated. Node dissection was therapeutically indicated in 45 patienta (16 percent) for whom no site of the primary melanoma was detected. The sites of the primary tumor were similar for each type of node dissection (Table I). The exception was a predominance of scalp tumors in patients with immediate therapeutic node dissection (40
372
Comparison ol Patknt Characterldlce by Type of Nodal Dtssectlon (values expressed as percentages)
Total patients (n) Site of primary lesion Scalp Ear Face Neck Upper trunk Multiple sites Unknown site Clinical nodal status Uninvolved Single, <3 cm Multiple, <3 cm Fixed Bilateral Biopsied Not described Involved nodes (n) 1 only 2-4 5-10 >lO Matted Not described Extranodal invasion Adjuvant nodal therapy
Elective
Dlssection of Nodes Unknown ImmeDeprimary diate layed Site
27
53
162
45
22 11 37 30 0 0 0
40 2 23 19 15 1 0
24 ;: 20 9 0 0
0 0 0 0 0 0 100
100 0 0 0 0 0 0
0 19 30 8 2 26 15
0 28 15 8 0 36 13
0 7 9 9 0 71 4
48 37 7 4 0 4 0 11
19 26 17 23 6 9 11 13
38 32 11 11 2 6 10 17
29 36 13 9 4 9 9 17
percent), as compared with those who had elective (22 percent) and delayed therapeutic (24 percent) node dissection. Of the patients who underwent an immediate therapeutic node dissection, the clinical nodal status (Table I) was recorded as a single node of less than 3 cm in 19 percent. In contrast, patients who underwent a delayed therapeutic node dissection had a higher incidence of a single involved node (28 percent). The number of histologically involved nodes was highest in patients who had immediate therapeutic node dissection and lowest in patients who underwent elective node dissection (Table I). One involved node was found in 48 percent of the patients with elective dissection, 38 percent of thoee with delayed therapeutic dissection, 29 percent of those with therapeutic dissection in whom the primary site was unknown, and 19 percent of those with immediate therapeutic dissection. Documented histologic extranodal invasion was absent in all patients who had elective node dissections, but occurred in a similar percentage of patients with immediate therapeutic dissection (11 percent), delayed therapeutic dissection (10 percent), and therapeutic dissection for metastases from an unknown primary site (9 percent) (Table I). Regional nodal recurrence after a neck dissection
Cervical Nodal M&stases From Cutaneous Melanoma
with a histologically involved node or nodes occurred in 42 patients (15 percent). Individually, factors such as sex and age of the patient, clinical stage at time of first evaluation at our hospital, site of the primary tumor, elective versus therapeutic node dissection, and type of adjuvant treatment after surgery did not influence nodal recurrence in the neck. The median age of the patients for whom nodal surgery failed was 43 years (range 17 to 83 years). Of these, 29 were male (14 percent) and 13 were female (15 percent). Regional therapy failed in 15 percent of the patients who initially presented at our hospital with cervical nodal metastases and underwent immediate neck dissection. This rate was not significantly higher than that for patients with disease confined to the primary site who had subsequent development of nodal metastases (9 percent). Of interest, but not of statistical significance, is the fact that 19 percent of the patients who had an elective node dissection exhibited recurrent disease in the neck dissection field compared with 17 percent of those who had immediate therapeutic dissection, 14 percent of those who had delayed therapeutic dissection, and 11 percent of those who had therapeutic dissection with an unknown primary site. The number of involved nodes did correlate with the rate of local therapy failure, however. Recurrent disease in the neck developed in 9 percent of the patients with only one histologically involved node. If 2 to 4 nodes were involved, 10 percent of the patients experienced a recurrence; however, if 5 to 10 nodes were involved or if more than 10 involved nodes were found, the recurrence rate increased to 22 percent and 26 percent, respectively. The highest recurrence rate (44 percent) was seen in patients whose involved nodes were described as matted. After neck dissection, in-transit disease occurred in 12 patients, consisting of primary tumors of the face in 7 patients, of the scalp in 3, of the upper trunk in 1, and of the neck in 1. Elective node dissections had been performed in four of these patients and delayed therapeutic dissection in eight. Nine of these patients had multiple involved nodes, and in seven, concomitant recurrent cervical nodal disease developed. Only 75 patients in this series (26 percent) remained free of disease after neck dissection. In 28 patients (10 percent), disease status at the time of death was unknown. The remaining 164 patients (64 percent) had documented distant metastases. Distant metastases were most common in patients with primary tumors of the scalp (77 percent), an ear (71 percent), the neck (63 percent), and the upper trunk (78 percent), and lowest in patients with primary tumors of the face (54 percent) or an unknown site (45 percent). There was no significant difference in the frequency of distant me&stases based on the timing of neck dissection in relation to treatment of
Volume 152, Octabm 1996
0
2
4
6
8
10
YeUS Fme 2. Swvlval rates from @e t&no of mMa dhsecthn to last foffow-up or desth. Cum A tdcates owratt auvtvaL Curva 6 Indicates mvfval wftf~d&ant m&stases after nods i%sssctkm
the primary tumor. The occurrence of distant metastases was lowest when only one histologically involved regional node was found (56 percent) and highest when there were multiple matted regional nodes (100 percent). Distant disease developed in 81 percent of patients with extranodal disease at the time of neck dissection. The firat sites of distant metastases were, in order of frequency, simultaneous multiple sites in 50 patients (27 percent), a lung in 41 (22 percent), the brain in 27 (15 percent), subcutaneous tissue in 23 (12 percent), the liver in 13 (7 percent), bone in 7 (4 percent), distant lymph nodes in 3 (2 percent), and the gastrointestinal tract in 3 (2 percent). In 17 patients (9 percent), the site of metastasis was known but not stated. The median length of survival from diagnosis of initial distantmetastases to the date of last follow-up or death was 21 months for the patients with distant lymph node involvement, 10 months for those with bone involvement, 5 months for those with lung and subcutaneous tissue involvement, 3 months for those with gastrointestinal tract involvement, and 2 months for those with brain or liver involvement or simultaneous multiple sites of metastases or nonspecified metastases. Only 2 of these 184 patients with distant metastasee remained free of disease after treatment (1 patient who underwent surgical resection of a solitary lung metastasis and 1 patient who received combination chemotherapy and immunotherapy for bilateral pulmonary lesions, with an 18 and 10 year follow-up, respectively). The 5 and 10 year survival rates for the entire group of 287 patients was 33 percent and 28 percent, respectively. Of the 184 patients (64 percent) who had subsequent development of documented distant me&stases after neck dissection, the 5 and 10 year survival rates decreased to 13 percent and 7 percent, respectively (Figure 2). No statistical difference in survival was detected based on any one of the following factors: age, sex, treatment of the primary tumor, clinical stage at first evaluation, or adjuvant treatment of the primary tumor or nodal
373
Singletaty el al
1.0 if
23
&cl .6 g
.4
d
A
3
n. ‘s
a .2
.O
c
0
2
4
6
a
10
Flpure 3. Survival raies fromthe the of node dbsectkm to lasf follow-up o.rdeath basedon the mmber oiInvolved nodea Curve A Indloates those paflents wtlh one Involved node: Curve 8, two
to lourInvolvednodes;CurveC, flve
or more Involved nodes; and
Curve D, matted nodes.
metastases. The length of survival of patients with a primary tumor of the head and neck region was also similar to that of patients with a primary tumor of the upper trunk (p = 0.792). The median survival from the onset of nodal metastases was 15 months for primary tumors of the scalp, 14 months for primary tumors of the ear, 19 months for those of the face, 18 months for those of the neck, and 20 months for those of the upper trunk. However, patients with primary tumors of unknown sites with cervical nodal metastases had significantly better lengths of survival (median 90 months) compared with patients with primary tumors of the head and neck (p = 0.002) or the upper trunk (p = 0.020). Survival was not affected by whether the indication for the node dissection was elective or therapeutic. Only 4 percent of the 162 patients who underwent delayed therapeutic node dissection had development of nodal metastases 24 months or more after treatment of the primary tumor. The median survival of this small number of patients was 24 months, compared with a median survival of 15 months for patients with nodal metastases that developed less than 24 months after the primary tumor was treated; however, this difference in survival was not statistically sigriificant. The numtir”bf involved nodes detected by neck dissection also influenced survival (Figure 3). Patients with only one involved node had a longer survival (median 39 months) than patients with two to four involved nodes (median 24 months, p = 0.021). A marked difference in length of survival was seen between patients with one involved node and those with five or more involved nodes (median 20 months, p = 0.0003) or matted nodes (median 8 months, p = 0.0001). No statistical difference in survival was detected between patients with two to four involved nodes versus five or more involved nodes (p = 0.099). The absence of extranodal disease significantly improved length of survival ((median 30 months, p = 0.0007) in contrast to that of
374
patients in whom melanoma was present in the connective tissue at dissection (median 11 months). The percentage of involved nodes (the number of involved nodes divided by the total number of nodes removed X 100 percent) was not predictive of survival. Less than 10 percent of the total number of cervical nodes removed were positive in 56 percent of the patients who died from disseminated melanoma. Regional failure after neck dissection also carried a grim prognosis. Of the 42 patients in whom nodal recurrence developed after a neck dissection, 26 percent survived 2 years and 7 percent survived 5 years. Of the 12 patients with in-transit disease, 31 percent survived 2 years and none survived 5 years. Thus, the three major factors that had an independent, statistically significant positive effect on the survival of patients with cervical nodal metastases were an unknown primary site, one histologically involved node, and the absence of extranodal disease.
The 33 percent cumulative 5 year survival rate of the 287 patients with cervical nodal metastases indicates that histologic evidence of involvement of nodes by malignant melanoma usually signifies disseminated disease. Even more dismal, the 13 percent 5 year survival of the 184 patients who had subsequent development of distant failure after regional node dissection testifies to the virulent behavior of melanoma. Although retrospective studies have shown inherent limitations, an understanding of the results of historical treatment may provide direction for future randomized, prospective clinical trials. Most important, the stratification of patients with nodal metastases by prognostic variables may identify certain subgroups of patients whose survival rates could be improved with regional or systemic adjuvant therapy. As confirmed by others [j-4], the dominant prognostic feature after the onset of regional nodal metastases is tumor burden, that is, the number of nodes affected by metastaaes. In this review, the best survival rate was seen in patients with tumor burden limited to a single histologically involved node. If the tumor extended extranodally, the incidence of both regional and distant failure increased dramatically. Age, sex, site of known primary disease, clinical stage at presentation, elective or therapeutic node dissection, and the time interval between treatment of the primary tumor and the development of nodal metastases were not predictive of survival after nodal disease was present. However, in patients with unknown primary sites, survival was significantly higher when compared with that of patients with known primary sites. Although some studies [5-71 have reported a similar favorable outcome for patients with unknown
The Amwlcan Journal ol Surgery
Cervical Nodal Metastases From Cutaneous Melanoma
primary sites, conflicting studies have shown a less favorable outcome or no change at all [a--101. Other possible prognostic criteria not evaluated in this review were the presence or absence of ulceration and Breslow’s tumor thickness of the primary tumor. In patients with histologically positive nodes, Balch et al [1] obtained a less than 15 percent 5 year survival rate in patients with ulcerated primary cutaneous melanoma in contrast to a 30 percent 5 year survival rate in patients with nonulcerative primary tumors. Day et al [II] found Breslow’s tumor thickness of the primary tumor to be predictive of survival in patients with clinically negative nodes and metastatic involvement limited to a positive node percentage of less than 20 percent. Only by clearly identifying comparable subgroups of patients with histologically involved nodes can valid conclusions be drawn as to optimal treatment. Summary A retrospective analysis with a minimum 10 year follow-up was performed on 287 patients who underwent radical or modified neck dissections with histologically involved regional nodal metastases from cutaneous malignant melanoma. The cumulative 5 year and 10 year survival rates calculated from the time of node dissection were 33 percent and 28 percent, respectively. Age and sex of the patient, site of known primary tumor, clinical stage at presentation, elective versus therapeutic node dissection, and time interval from the treatment of the primary tumor to node dissection did not independently affect survival. However, an unknown site of primary disease, the presence of only one histologi-
volume 152, octobw lB86
cally involved node, and the absence of extranodal tumor invasion at the time of node dissection were statistically significant individual prognostic factors for an improved survival rate. References 1. Balch CM, Soong SJ, Murad TM, lngalls AL, Maddox WA. A multifactorial analysis of melanoma. Ill. Prognostic factors in melanoma patients with lymph node metsstases (stage Ill). Ann Surg 1981;193:377-88. 2. Cohen MH, Ketcham AS, Felix FI, et al. Prognostic factors in patients undergoing lymphadenectomy for malignant melanoma. Ann Surg 1977;188:835-42. 3. Fortner JG, Woodruff J, Schottenfeld D, MacLean B. Blostatistical basis of elective node dissection for malignant melanoma. Ann Surg 1977;186:101-3. 4. Karakousis CP, Seediq MK, Moore R. Prognostic value of lymph node dissection in malignant melanoma. Arch Surg 1980;115:719-22. 5. Baab GH, McBride CM. Malignant melanoma; the patient with an unknown site of primary origin. Arch Surg 1975;llO: 896-900. 6. Lopez R, Holyoke ED, Moore RH, Karakousis CP. Malignant melanoma with unknown primary site. J Surg Oncol 1982; 19:151-4. 7. Santini H, Byers RM, Wolf PF. Metastatic melanoma to cervical and parotid nodes from an unknown primary site. Am J Surg 1985;150:510-2. 8. Milton GW, Shaw HM, McCarthy WH. Occult primary malignant melanoma, factors influencina survival. Br J Surg 1977;64:805-8. 9. Das Gupta T, Bowden L, Berg J. Malignant melanoma of unknown primary origin. Surg Gynecol Obstet 1963;117: 341-50. 10. Giuliano AE, Moseley HS, Morton DC. Clinical aspects of unknown primary melanomas. Ann Surg 1980;191:98104. 11. Day CL, Sober AJ, Lew RA, et al. Malignant melanoma patients with positive nodes and relatively good prognoses. Cancer 1981;47:955-62.
375