SEARCH

SEARCH BY CITATION

Keywords:

  • desmoplastic;
  • melanoma;
  • radiation;
  • radiotherapy;
  • control;
  • local neoplasm recurrence

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

Desmoplastic melanoma may have a high risk of local recurrence after wide excision. The authors hypothesized that adjuvant radiotherapy (RT) would improve local control in patients with desmoplastic melanoma, resulting in at least a 10% absolute decrease in local recurrence rate.

METHODS

A total of 277 patients from 1989 through 2010 who were treated for nonmetastatic desmoplastic melanoma by surgery with or without adjuvant RT were reviewed. Clinicopathologic and treatment variables were assessed with regard to their role in local control.

RESULTS

A total of 113 patients (40.8%) received adjuvant RT. After a median follow-up of 43.1 months, adjuvant RT was found to be independently associated with improved local control on multivariable analysis (hazards ratio, 0.15; 95% confidence interval, 0.06-0.39 [P < .001]). Among 35 patients with positive resection margins, 14% who received RT developed a local recurrence versus 54% who did not (P = .004). In patients with negative resection margins, there was a trend (P = .09) toward improved local control with RT. In patients with negative resection margins and traditionally high-risk features, including a head and neck tumor location, a Breslow depth > 4 mm, or a Clark level V tumor, RT was found to significantly improve local control (P < .05). The data from the current study would suggest that patients who would be good candidates for omitting RT included those with negative resection margins, a Breslow depth ≤ 4 mm, and either no perineural invasion present or a non-head and neck tumor location.

CONCLUSIONS

RT for desmoplastic melanoma was independently associated with improved local control. Patients with positive resection margins or deeper tumors appeared to benefit the most from RT, whereas selected low-risk patients can safely omit RT. Cancer 2014;120:1369–1378. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Desmoplastic melanoma is a rare variant of melanoma that is reported to account for < 4% of all primary cutaneous melanomas.[1] As described by Busam et al,[2] desmoplastic melanoma consists of fusiform melanocytes dispersed in a prominent collagenous stroma. Furthermore, desmoplastic melanoma has been divided into a “pure” subtype, with > 90% desmoplasia, and a “combined” or mixed subtype, with ≤ 90% but > 10% desmoplasia.[2, 3] Desmoplastic melanoma frequently exhibits a less characteristic appearance than other cutaneous melanomas and is often amelanotic, potentially resulting in diagnostic delays and increased tumor thickness at the time of diagnosis.[4, 5] In many series, desmoplastic melanoma has been reported to have a higher local recurrence rate after surgical removal compared with other melanomas. Analysis of patterns of failure demonstrates a broad range of reported local recurrence rates in patients with desmoplastic melanoma, from 4% to 48%, after excision alone.[1, 6-10] Potential risk factors for local recurrence include a tumor location on the head and neck, Clark level V, a Breslow depth > 4 mm, neurotropism/perineural invasion, and histologically positive surgical resection margins.[1, 4, 11] We and others have hypothesized that adjuvant radiotherapy (RT) would improve rates of local control after wide excision.[6, 11, 12] Therefore, we reviewed our experience with RT in patients with desmoplastic melanoma to analyze the impact of adjuvant postoperative RT on local recurrence rates. We hypothesized that an absolute reduction in local and/or locoregional recurrence of at least 10% would justify a routine recommendation for treatment.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Patients and Histopathologic Analysis

After obtaining Institutional Review Board approval, a retrospective review from 1989 through 2010 identified 332 patients diagnosed with desmoplastic melanoma. After excluding patients who presented with distant metastasis or locally recurrent disease, those who declined surgery or who received prior RT to the area, and those with unavailable treatment records, 277 patients remained for analysis. All patients had American Joint Committee on Cancer[13] stage I to III melanoma (any T, any N, M0: any tumor size with or without ulceration with or without regional lymph node metastasis, no distant metastasis [M0]), with no clinical, radiologic, or pathologic evidence of metastasis beyond regional lymph nodes. Demographic, histopathologic, RT, and outcome data were reviewed.

Patients were treated with wide excision with or without sentinel lymph node biopsy. The majority of patients underwent wide excision with a 2-cm margin and a minimum margin of 1 cm in regions of the body with complex neurovascular anatomy, including the head and neck. Patients with negative resection margins either at the time of initial wide excision or at reexcision were considered to have negative resection margins. Close resection margins were considered to be negative. Patients who underwent initial excisional biopsy with positive resection margins followed by reexcision to negative resection margins were classified as having negative resection margins. Patients with a positive lymph node, whether microscopic or macroscopic, were routinely offered a regional lymph node dissection.

Primary tumor characteristics were assessed and rereviewed by a single dermatopathologist (J.L.M.) in cases with available tissue (64% of all cases). The Memorial Sloan-Kettering Cancer Center classification system was used to classify desmoplastic melanoma as pure or mixed, as first described by Busam et al and later modified by Hawkins et al.[2, 3] Spindle cell melanomas with ≥ 90% desmoplasia were defined as “pure,” whereas melanomas with < 90% but > 10% desmoplasia were defined as “mixed.” Lymph nodes were evaluated by serial sectioning and through the use of hematoxylin and eosin-stained sections. S-100 and melan-A immunohistochemistry were used in the evaluation of all sentinel lymph nodes.

Radiotherapy

The majority of treated patients (58%) received adjuvant RT at Moffitt Cancer Center, whereas the remainder received treatment at outside facilities. Patients who received RT were treated with either electrons or photons to the postoperative primary tumor bed, with or without the regional lymph nodes being included as indicated. The primary tumor site was targeted to encompass the scar plus approximately 2 cm to 4 cm. The majority of patients who received adjuvant RT were treated with 1 of 2 regimens. Most were treated to a total dose of 30 grays (Gy) in 600-centigray (cGy) fractions given twice per week for 2.5 weeks as described by Ang et al.[14] When electrons were delivered, the electron energy was appropriate for the depth of the tumor and surgical bed (typically 6 megaelectron volts [MeV]-12 MeV) and prescribed to D-maximum (point dose). A bolus was included to prevent underdosing of the tumor bed when photons and low-energy electrons were used (6 MeV-9 MeV). The remaining patients received a total dose of 59.4 Gy to 68 Gy in once-daily fractions of 180 cGy to 200 cGy. With photon treatment, computed tomography-based treatment planning was used to limit the volume of normal tissue in the radiation field, for both dosimetric calculations and radiation plan evaluation. Tissue inhomogeneity did not exceed a 5% to 10% increase or decrease in dose prescription across the target volume. The regional lymphatics were generally included in the treatment field if a lymph node measured > 3 cm, if there was extranodal extension present, or if ≥ 3 lymph nodes were pathologically positive for malignancy.

Statistical Analysis

The primary endpoint was local control, defined as freedom from disease recurrence within the radiated field of the primary tumor bed. The secondary endpoint was locoregional control, defined as freedom from disease at the primary tumor bed, in-transit region, and regional lymph nodes. We also evaluated the rate of distant metastasis, defined as the rate of tumor spread beyond the regional lymph nodes. Clinicopathologic factors associated with survival will be addressed in a subsequent article. The primary and secondary outcomes were assessed first among all patients, then among only those patients with negative resection margins, and finally among only patients with negative resection margins who did not receive adjuvant RT. Patient subgroups with negative resection margins were then assessed with respect to each potentially high-risk variable, including age, tumor location, Breslow depth > 4 mm, Clark level V, pure versus mixed histology, ulceration, and perineural invasion to determine which subgroups might significantly benefit from RT. We used a 10% absolute improvement in local or locoregional control as a cutoff because we believed that an absolute improvement of ≥ 10% in in-field disease control with RT would provide a clinically meaningful threshold for deciding whether to deliver adjuvant RT and would outweigh an estimated incidence of ≤ 5% for serious acute and late side effects of RT to the soft tissues and/or lymph node basins involved. We based the estimated ≤ 5% side effects with RT on findings by Ang et al,[14] who reported severe acute radiation side effects, including moist desquamation and confluent mucositis, in < 5% of patients who underwent postoperative electron RT for melanoma of the head and neck; only 2% of patients in the study experienced severe, nonfatal, late radiation complications. We present both absolute differences in subgroups and Kaplan-Meier estimates using the log-rank test, because linear rank tests can be inaccurate when sample sizes are small and/or unbalanced.[15]

We also attempted to identify subgroups of patients in whom RT might be safely omitted. To do so, we sought subgroups that not only demonstrated a lack of significant benefit with RT (P > 0.05), but also had an absolute difference of ≤ 5% in local control with RT, at which point the side effects might outweigh the benefits of RT.

Statistical analysis was performed using SPSS statistical software (version 21.0; IBM, Chicago, Ill). The Wilcoxon rank sum test was used to compare continuous variables, whereas the Pearson chi-square test or Fisher exact test were used to compare categorical variables as appropriate, with an added category for variables with unknown results. Actuarial rates of local control and locoregional control were calculated using the Kaplan-Meier method and differences in rates based on individual variables were assessed with the log-rank test. All clinical, histopathologic, and treatment variables were added to Cox multivariable regression models. Variables with unknown quantities < 5% of the total count were included in multivariable analyses without their unknowns. Continuous variables were split using clinically meaningful cutpoints. All tests were 2-sided and an α (type I) error of ≤ .05 was considered to be statistically significant.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Patient and Tumor Characteristics

The median follow-up was 43.1 months, the median age of the patients was 68 years (range, 16 years-96 years), and the median Breslow depth was 3.9 mm (range, 0.5 mm-35 mm). There were more men than women (male:female ratio of 2.7:1) with desmoplastic melanoma, and men were more likely than women to have desmoplastic melanoma at an age > 70 years (47% vs 28%), have head and neck tumors (61% vs 39%), and have ulcerated tumors (28% vs 5%).

RT Characteristics

Patient, tumor, and treatment characteristics for all 277 evaluable patients are shown by treatment received in Table 1. In total, 113 patients (41%) received postoperative RT. The majority of patients (74%) were treated with en face electrons. Before 2005, 81% of patients were treated with a hypofractionated regimen of 30 Gy in 5 fractions, whereas after 2005, 90% of patients were treated with conventional fractionation using fractions of 180 cGy to 200 cGy to a total dose ranging from 59.4 Gy to 68 Gy. There was no significant difference in actuarial rates of local control between patients who received the hypofractionated regimen versus those treated with the conventional regimen (log-rank P = .92). On univariable regression analysis (Table 1), adjuvant RT was used significantly more frequently for patients with tumors of Clark level V, a head and neck tumor location, > 4-mm Breslow thickness, and, for those with pure desmoplastic melanoma, perineural invasion or positive surgical resection margins. Using a logistic regression model, patients with head and neck tumors (odds ratio [OR], 2.4; 95% confidence interval [95% CI], 1.4-4.3 [P = .003]), Clark level V tumors (OR, 2.3; 95% CI, 1.1-4.5 [P = .02]), or tumors measuring > 4 mm in thickness (OR, 2.9; 95% CI, 1.5-5.7 [P = .002]) were significantly more likely to have received adjuvant RT.

Table 1. Patient, Tumor, and Treatment Characteristics and Associations With Adjuvant RT (n = 277)
Characteristics No RT % (No.)RT % (No.)P
  1. Abbreviations: CLND, completion lymph node dissection; Gy, Grays; LN, lymph node; PNI, perineural invasion; RT, radiotherapy; SD, standard deviation; SLNB, sentinel lymph node biopsy; TLND, therapeutic lymph node dissection.

Mean age ± SD, y 66.0 ± 15.365.3 ± 13.9.58
Median age (range), y 69 (16-96)68 (25-89) 
Age, y>7043 (70)41 (46).81
 ≤7057 (94)59 (67) 
SexMale70 (114)78 (88).13
 Female30 (50)22 (25) 
LocationHead and neck44 (72)71 (80)<.001
 Other56 (92)29 (33) 
Mean Breslow depth ± SD, mm 4.1 ± 3.56.4 ± 4.4<.001
Median Breslow depth (range), mm 2.9 (0.5-18.0)5.0 (1.0-35.0) 
Breslow depth, mm>433 (54)65 (74)<.001
 ≤466 (108)32 (36) 
 Unknown1 (2)3 (3) 
Clark levelV33 (54)68 (77)<.001
 III to IV63 (104)27 (31) 
 Unknown4 (6)4 (5) 
HistologyPure28 (46)44 (50).01
 Mixed30 (50)28 (32) 
 Unknown41 (68)27 (31) 
UlcerationPresent14 (23)24 (27).1
 Absent63 (104)58 (66) 
 Unknown23 (37)18 (20) 
PNIPresent43 (70)58 (65).005
 Absent30 (50)31 (35) 
 Unknown27 (44)12 (13) 
Resection margin statusPositive8 (13)19 (22).01
 Negative90 (148)79 (89) 
 Unknown2 (3)2 (2) 
SLNBYes67 (110)74 (84).23
 No33 (54)26 (29) 
Positive SLNB with CLNDYes9 (15)10 (11).95
 No90 (148)89 (101) 
 Unknown1 (1)1 (1) 
TLNDYes1 (2)3 (3).4
 No99 (162)97 (110) 
Involved regional LNsYes11 (18)13 (15).58
 No89 (146)87 (98) 
RT total dose, Gy3045 (51)
 50-68 Gy39 (44) 
 Other4 (4) 
 Unknown12 (14) 
RT fractions544 (50)
 25-3635 (39) 
 Other6 (7) 
 Unknown15 (17) 
RT deliveryElectrons74 (84)
 Photons12 (13) 
 Unknown14 (16) 
LN RTYes5 (6)
 No85 (98) 
 Unknown10 (11) 

Regional Lymph Node Treatment Characteristics

Clinical and treatment characteristics of regional lymph nodes are shown in Table 1. A total of 194 patients (70%) underwent sentinel lymph node biopsy, with a positive lymph node found in 28 patients (14%). Of these 28 patients, 26 (93%) underwent a completion lymphadenectomy, whereas it is unknown whether lymphadenectomy was performed in 2 patients. Four of the 26 patients (15%) who underwent completion lymphadenectomy had additional positive lymph nodes (range, 1 lymph node-6 lymph nodes). Five patients (2%) presented with clinically palpable and histologically positive lymph nodes, and all underwent a therapeutic lymphadenectomy. In total, 33 patients (12%) had pathologically proven regional lymph node involvement, including 11 patients with pure desmoplastic melanoma (12% of all patients with pure desmoplastic melanoma).

Adjuvant local plus regional RT was delivered after lymphadenectomy in only 6 patients, 3 of whom had a positive sentinel lymph node and 3 of whom had macroscopic lymph node disease. There were no significant differences noted between the patients who did and those who did not receive local ± regional RT in terms of lymph node staging or involvement.

Local Control

In total, 36 of 277 patients (13%) failed locally; the median time to local failure was 14 months (range, 2 months-113 months). Variables found to be associated with local control among all patients with desmoplastic melanoma are shown in Table 2. Adjuvant RT was associated with improved local control using log-rank analysis (P = .02) and Cox multivariable analysis (hazards ratio [HR], 0.15; 95% CI, 0.06-0.39 [P < .001]). Factors found to be significantly associated with poorer local control included male sex (HR, 3.8; 95% CI, 1.3-11.2 [P = .01]), Clark level V tumors (HR, 2.3; 95% CI, 1.0-4.9 [P = .04]), and positive surgical resection margins (HR, 6.6; 95% CI, 2.8-15.7 [P < .001]). Twenty-eight of 164 patients (17%) who did not receive RT developed a local recurrence, compared with only 8 of 113 patients (7%) who received postoperative RT. The 1-year and 5-year actuarial local control rates were 96% and 95%, respectively, for patients treated with adjuvant RT versus 91% and 76%, respectively, for unirradiated patients (log-rank P = .015) (Fig. 1 Top).

Table 2. Variables Associated With Local Control and Locoregional Control Among All Patients with Desmoplastic Melanoma
Variable    Local Control  Locoregional Control
  KM 5-Year LC, %Log-RankCox Multivariable RegressionCrude LRC, %KM 5-Year LRC, %Log-RankCox Multivariable Regression
% (No.)Crude LC, %PHR (95% CI)PPHR (95% CI)P
  1. Abbreviations: 95% CI, 95% confidence interval; HR, hazards ratio; KM, Kaplan-Meier; LC, local control; LRC, locoregional control; PNI, perineural invasion; RT, radiotherapy.

Age, y>7042 (116)8378.0037568.0022.4 (1.3-4.2).004
 ≤7058 (161)9088   8483   
SexMale73 (202)8580.033.8 (1.3-11.2).017874.06
 Female27 (75)9393   8786   
LocationHead and neck55 (152)8583.247877.44
 Other45 (125)9085   8278   
Breslow depth, mm>447 (128)8682.237673.022.5 (1.4-4.7).003
 ≤453 (144)9087   8583   
Clark levelV49 (131)8479.082.3 (1.0-4.9).047673.04
 III-IV51 (135)9088   8481   
HistologyPure54 (96)8481.847775.59
 Mixed46 (82)8477   7468   
UlcerationPresent23 (50)8681.687268.07
 Absent77 (170)8885   8381   
PNIPresent61 (135)8480.387471.15
 Absent39 (85)8984   8582   
Resection marginsPositive13 (35)7166<.0016.6 (2.8-15.7)<.0016053<.0014.4 (2.2-8.7)<.001
 Negative87 (237)9087   8481   
RTYes41 (113)9395.020.15 (0.06-0.39)<.0018789.020.20 (0.10-0.41)<.001
 No59 (164)8376   7669   
image

Figure 1. Kaplan-Meier plots for local control are shown in patients with desmoplastic melanoma (Top) who were treated with or without adjuvant radiotherapy (RT) (277 patients) and (Bottom) who were with or without positive resection margins (272 patients; excluding 5 patients with unknown resection margin status). Post-op indicates postoperative.

Download figure to PowerPoint

Thirty-five patients (13%) had a positive resection margin, 237 patients (86%) had a negative resection margin, and 5 patients had unknown margin status. Ten of 35 patients with positive resection margins (29%) failed locally compared with 24 of 237 patients with negative resection margins (10%; log-rank P < .001) (Fig. 1 Bottom). The largest difference in local control between patients who received RT compared with those who did not was observed for patients with positive resection margins.

A flow chart of RT delivered by margin status and location is shown in Figure 2. Among 35 patients with positive resection margins, 3 of 22 patients (14%) who received RT developed a local recurrence compared with 7 of 13 (54%) of those who did not (P = .003). The benefit of RT in patients with negative resection margins was not as pronounced; 5 of 89 radiated patients (6%) failed locally versus 19 of 148 unirradiated patients (13%) (P = .09). Among patients with negative resection margins, the only factor found to be significantly associated with poorer local control on multivariable regression analysis was age > 70 years (HR, 2.4; 95%CI, 1.1-5.4 [P = .04]). However, there were subsets of patients with negative resection margins who did appear to benefit from adjuvant RT. Patients with negative resection margins and high-risk features, including a head and neck location, a Breslow depth > 4 mm, or a Clark level V tumor, had significantly improved local control with the use of RT on log-rank analysis, and a ≥ 10% difference in absolute rates of local control (Table 3).

Table 3. Univariable Analysis of Outcomes With and Without RT in Resection Margin-Negative Cases (n = 237), and Patients with Traditionally Higher-Risk Features
   Local ControlLocoregional Control
  % (No.)Crude, %KM 5-Year, %Log-Rank PCrude, %KM 5-Year, %Log-Rank P
  1. Abbreviations: KM, Kaplan-Meier, PNI, perineural invasion; RT, radiotherapy.

Age >70 yRT34 (32)9190.348483.33
n = 93None66 (61)8579 7970 
Male sexRT40 (68)9494.069192.02
n = 170None60 (102)8477 7770 
Head and neck locationRT48 (57)9596.039090.08
n = 120None53 (63)8176 7672 
Breslow depth >4 mmRT55 (57)9595.039091.002
n = 103None45 (46)8375 6762 
Clark level VRT58 (60)9795.0019291<.001
n = 104None42 (44)7765 6658 
Pure desmoplasiaRT52 (43)9394.068889.03
n = 83None48 (40)8073 7365 
Mixed histologyRT32 (21)9195.688190.63
n = 65None68 (44)8669 7762 
PNIRT45 (50)9292.128686.04
n = 110None55 (60)8274 7062 
image

Figure 2. Breakdown of patients with desmoplastic melanoma (DM) treated with or without adjuvant radiotherapy and their corresponding Kaplan-Meier (KM) 5-year actuarial local control (LC) and crude LC rates with log-rank P values is shown. Patients with negative resection margins, a non-head and neck tumor location, and a Breslow depth ≤ 4 mm had a < 5% benefit with radiotherapy.

Download figure to PowerPoint

In an attempt to clarify what factors were predictive of local recurrence in the absence of RT, 148 patients with negative resection margins who did not receive RT were assessed. In this subset, only Clark level V tumors were found to be significantly associated with poorer local control on multivariable analysis (HR, 3.7; 95% CI, 1.5-9.5 [P = .006]). There was also a trend toward statistical significance among patients aged > 70 years and those with head and neck tumors on univariable analysis (log-rank P = .10 and P = .06, respectively). Patients with negative resection margins who did not receive RT in our analysis were generally those with very few high-risk features and, as a result, had few instances of disease recurrence.

Patients with desmoplastic melanoma who might not benefit from RT were then assessed. Two subsets of patients demonstrated an absolute improvement of ≤ 5% in local control with RT, including patients with negative resection margins, a non-head and neck tumor location, and a Breslow depth ≤ 4 mm (92% local control with RT [12 patients] vs 96% without [54 patients]; P = .63) (Fig. 2), and patients with negative resection margins, a Breslow depth ≤ 4 mm, and no perineural invasion (100% local control with RT [11 patients] and 97% without [36 patients]; P = .58) (data not shown).

Locoregional Control

Variables associated with locoregional control are shown in Table 2. In total, only 21 of 264 patients (8%; 7 with pure, 10 with mixed, and 4 with unknown histology) developed a regional disease recurrence, alone or as part of a local or distant failure. Patients treated with RT were found to have significantly improved locoregional control on univariable (P = .02) and multivariable (HR, 0.20; 95% CI, 0.10-0.40) (P < .001) analysis. Forty of 164 patients (24%) who did not receive local/regional RT developed a locoregional failure versus 15 of 113 patients (13%) who did. Other variables found to be significantly associated with poorer locoregional control included age > 70 years (HR, 2.4; 95% CI, 1.3-4.2 [P = .003]), Breslow depth > 4 mm (HR, 2.5; 95% CI, 1.4-4.7 [P = .003]), and positive resection margins (HR, 4.6; 95% CI, 2.3-9.1 [P < .001]). Of 6 patients who received adjuvant locoregional RT for positive lymph nodes, 1 (17%) developed a regional recurrence, versus 2 of 27 patients (7%) with positive lymph nodes who did not receive regional RT. The use of regional lymph node RT was significantly associated with positive resection margins (OR, 7.0; 95% CI, 1.4-36.2 [P = .02]).

In patients with positive resection margins, 23% had a locoregional recurrence with RT versus 69% without (log-rank P = .002). In contrast, among patients with negative resection margins, 10% experienced a locoregional recurrence with RT versus 20% without (P = .06). On multivariable regression analysis, patient age > 70 years, Breslow depth > 4 mm, and no RT were all found to be associated with poorer locoregional control in patients with negative resection margins (P < .05 for each). Among high-risk subsets of patients with negative resection margins, variables found to be associated with significantly improved locoregional control with adjuvant RT included male sex and those patients with deeper tumors, pure desmoplasia, or perineural invasion (Table 3).

Distant Metastasis Rate and Salvage Surgery

Sixty-three of 277 patients (23%) eventually developed distant metastases with a median time from wide excision of 17 months (range, 2 months-121 months). Of 36 patients experiencing local failures, 7 were surgically salvaged, 10 developed multiple local recurrences and eventually died of desmoplastic melanoma or related complications, 2 presented with simultaneous distant metastases, and 16 went on to develop distant metastases (median time from local failure to distant metastasis, 20 months; range, 4 months-101 months), whereas the final patient was lost to follow-up.

Toxicity of Radiation

Acute and long-term side effects from RT were as expected. Common acute side effects included skin erythema, pain, and fatigue, whereas long-term side effects included skin fibrosis, telangiectasias, and skin pigment changes. Given the retrospective nature of the current study, a validated tool for grading toxicity was not clearly documented for all patients.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

The results of the current study demonstrate that adjuvant RT after surgery for desmoplastic melanoma is significantly associated with improved local and locoregional control. Consistent with previous reports, a majority of patients with desmoplastic melanoma were older men with a head and neck tumor location. The median Breslow depth in this series was 3.9 mm, similar to other desmoplastic melanoma series,[2-5, 8, 10, 12] but higher than reports from the group in New South Wales.[1, 7] The relatively thick median Breslow depth in the current series may be a result of the predominantly older male population, who had more locally advanced tumors than women. Most patients treated with RT in the current study had tumors with multiple adverse prognostic features, including a head and neck location, Breslow depth > 4 mm, and Clark level V. These factors have been repeatedly associated with worse local control, recurrence-free survival, and overall survival in patients with desmoplastic melanoma.[1, 2, 4, 6-8, 11, 16, 17] Because of the multiple adverse risk features in patients treated with RT, we performed a multivariable regression analysis to attempt to control for confounding factors with respect to local and locoregional control.

The overall local recurrence rate of 13% in the current study falls well within the range reported in other published series (4%-48%).[1, 6-10, 16, 18] Positive resection margins placed patients at a higher risk of local recurrence than any other variable. Positive resection margins have previously been shown to be associated with increased rates of local recurrence with desmoplastic melanoma.[4, 5, 10, 11, 16] In this series, there was a striking (40%) absolute reduction in local recurrence among patients with positive resection margins who received RT (3 of 22 patients; 14%) compared with those who did not (7 of 13 patients; 54%). To the best of our knowledge, this is among the strongest evidence to date supporting the use of adjuvant RT in patients with positive resection margins. To get a better idea of who might benefit from RT with negative resection margins, we assessed that group separately.

Among all patients with desmoplastic melanoma and negative resection margins, there was a trend toward improved local control with the use of RT (P = .09). Because of the trend, we assessed traditionally higher-risk patient subgroups to determine who might benefit from RT with negative resection margins. As shown in Table 3, patients with negative resection margins and a head and neck tumor location or deeper tumors were found on univariable regression analysis to significantly benefit from RT. There was also a trend toward improved local control in patients with negative resection margins and pure desmoplasia (P = .06) or male sex (P = .06). We should point out that patients with perineural invasion did not significantly benefit from RT (P = .12), although the number of patients in the group was fairly small (50 patients in the radiated vs 60 patients in the nonirradiated group), and the Kaplan-Meier curves did separate over time, with a 5-year actuarial local control difference of 18% (92% in those treated with RT and 74% in those treated without) (data not shown). As a result, patients with perineural invasion might represent a group that significant benefits from adjuvant RT, given more power to detect a difference.

In an attempt to clarify who might not require adjuvant RT, subgroups of patients with lower-risk features were assessed. Because the number of patients in some subgroups became quite small (for example, there were 66 patients who had a non-head and neck tumor location and a Breslow depth ≤ 4 mm, with only 12 of those individuals having received RT), the power to detect a significant difference on log-rank analysis also diminished. With that in mind, we identified 2 possible subgroups of low-risk patients in whom the log-rank P value was > .05 and the absolute difference in local control was ≤ 5%, including patients with negative resection margins, a non-head and neck tumor location, and a Breslow depth ≤ 4 mm (P = .63) (Fig. 3), and patients with negative resection margins plus a Breslow depth ≤ 4 mm and no perineural invasion present (P = .58) (data not shown). The data from the current study suggest these 2 subgroups could safely be spared adjuvant RT.

Three previous studies have suggested a possible benefit in local control with RT in patients with desmoplastic melanoma; however, all were underpowered to demonstrate a clear benefit.[6, 11] Chen et al[11] retrospectively reviewed 128 patients with desmoplastic melanoma, 27 of whom received RT after undergoing surgical excision. Similar to our experience, tumors treated with RT were more likely to have poorer prognostic features, such as increased Breslow thickness and Clark level, ulceration, or close or positive resection margins. Of the 27 patients treated with RT, 2 (7%) developed local recurrence compared with 6 of 101 patients (6%) who did not receive RT. The absolute percentages of patients who developed local recurrence were similar despite RT being used more often among patients with aggressive lesions. Vongtama et al[6] also reviewed 44 patients with desmoplastic melanoma, 21 of whom developed local recurrence. Of 14 patients with a local recurrence who were treated with salvage surgery and RT, none of the tumors recurred locally, compared with 4 of 7 patients treated with surgery alone. A third study by Foote et al[12] reviewed 24 patients with melanoma who were treated with RT, 22 of whom had desmoplastic melanoma, and demonstrated a promising 3-year, in-field, recurrence-free survival rate of 91%. Although these studies suggest promising results with RT in patients with desmoplastic melanoma, they were too small to demonstrate a clear benefit. In contrast, the current study, which involved 277 patients, demonstrated a clear improvement in local control with the use of RT.

Among all patients with desmoplastic melanoma in the current series, adjuvant RT was associated with improved locoregional control on both univariable and multivariable analysis; however, this benefit appears to be primarily from improved local control. Because desmoplastic melanoma spreads to the lymph nodes in only a minority of cases (12% in the current series and 14% in a previous analysis performed by our group among patients with desmoplastic melanoma who underwent sentinel lymph node biopsy[19]), regional RT was rarely used. As a result, the role of local plus regional lymph node RT could not be adequately addressed.

The current study was limited by its single-institution, retrospective nature, and by the long time period from 1989 to 2009. During this timeframe, treatment protocols and pathologic evaluation varied, and there were improvements in radiation technology and treatment planning. Patients were treated with 2 distinct RT regimens, which shifted from a hypofractionated approach to a more conventional fractionation scheme. We found acceptable rates of acute and late toxicity with both the standard fractionation and the hypofractionation regimens. Our institutional practice pattern changed after completion of the Florida Melanoma Trial[20] and given concerns regarding long-term toxicity with a high dose-per-fraction regimen to a large volume of normal tissue. Although we have a preference for standard fractionation, we still use the high dose-per-fraction regimen in selected cases. When hypofractionation is considered, we take into account the volume of tissue to be irradiated, wound healing, the anatomic location of the scar, the age of the patient, and the ease of treatment among other factors. Despite the shift in RT schedules from a hypofractionated approach to a conventional approach, it did not appear to affect local control rates.

The use of adjuvant RT after wide excision is significantly associated with improved local control and locoregional control. Therefore, RT has a role in patients with desmoplastic melanoma with high-risk clinical or pathologic features. Patients with positive resection margins and deeper tumors appeared to benefit the most from RT, whereas patients with negative resection margins, a Breslow depth ≤ 4 mm, and either no perineural invasion or a non-head and neck tumor location could safely omit RT.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Dr. Caudell has acted as a consultant for Bristol-Myers Squibb. Dr. Zager has acted as a consultant/member of the Scientific Advisory Board and received a grant from Delcath Systems Inc, has acted as a consultant/member of the Scientific Advisory Board for IGEA and Amgen, and has acted as a consultant/member of the Advisory Board and received a grant from LifeCell. Dr. Messina has acted as a consultant for GlaxoSmithKline and the DURECT Corporation. Dr. Sondak has acted as a consultant/member of the Advisory Board for Merck, Navidea, and Provectus.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES