SEARCH

SEARCH BY CITATION

Desmoplastic melanoma (DM) represents a small (approximately 4%), but distinct subset of cutaneous malignant melanoma. As a consequence of its rarity, virtually all treatment information is gleaned from retrospective studies. Current recommendations regarding the use of adjuvant radiotherapy (XRT) in this disease are based on published data consisting of a relatively modest total number of patients.[1-3] In this issue of Cancer, 2 large retrospective series focusing on the role of adjuvant external-beam XRT in the management of patients with DM are presented. Guadagnolo et al[4] report their experience with 130 patients treated at The University of Texas MD Anderson Cancer Center (MDACC) and Strom et al[5] present their data regarding 277 patients whose care was managed through Moffitt Cancer Center. At each institution, common treatment approaches were used, but essentially all final therapy decisions were made at the discretion of the individual physicians. Both groups reported prognostic factors for their entire patient populations, but also analyzed their data to identify the selected individuals who are most likely to benefit from adjuvant XRT or those who do not require it. The combined information contained in these 2 studies, which adds another 400 patients with DM to the published literature, helps to solidify treatment recommendations regarding adjuvant XRT and also points to areas remaining to be clarified.

DM, which was first identified in 1971,[6] was initially believed to behave more aggressively than other melanoma subtypes. In part, this sense derived from the finding that DMs were being diagnosed at a more advanced stage.[7] The growth pattern and clinical appearance of DM can make them more difficult to detect, and therefore they often have progressed further than other melanomas before being identified as a malignant lesion requiring surgical resection. However, it does not appear to be the case that DM represents a more aggressive histology. The uncertainty regarding the clinical behavior of DMs was addressed in a matched-pair analysis comparing DMs with other cutaneous melanomas.[8] This study demonstrated that, stage for stage and depth for depth, the 2 groups were similar with respect to overall survival. Although overall survival was comparable, there are extensive data demonstrating that the intrinsic biology of DM causes an increased propensity for local disease recurrence and a relative decrease in regional lymph node metastases. Initially, the relative importance of this aspect of their behavior may have been masked by their growth patterns, which often led to suboptimal resections resulting in inadequate surgical margins leading to a decrease in local control (LC).

The pathologic classification system for DM is evolving and remains to be fully incorporated into overall management. Busam et al[9] have identified 2 classes of DMs: “pure” and “mixed” based on the percentage of DM observed in a pathologic specimen (pure is > 90% vs mixed at > 10%, but < 90%). In studies separating out these 2 entities, the patients with pure DM demonstrated a characteristic biological behavior, compared with patients with mixed tumors, which generally behave more like the overall group of cutaneous melanomas. For example, in a study of lymph node involvement, Pawlik et al[10] demonstrated that patients with pure DMs had a much lower incidence of positive sentinel lymph node biopsies (SLNBs) compared with patients with all other melanomas, whereas patients with mixed DMs had results similar to those of patients with other melanomas. In the studies under consideration, Strom et al had a pathologist re-review 64% of their samples with respect to these criteria, whereas Guadagnolo et al relied on the original pathology reports to put patients within the framework of the Busam classification system.[4, 5]

DM is a malignancy presenting a predominantly local control issue and thus surgical resection and XRT are the basic elements of definitive treatment. With regard to the global question of whether patients with DM receive adjuvant XRT, the answer from both articles is: yes! If we consider the entire population of each study, the patients treated with postoperative XRT demonstrated improved LC compared with those treated with surgery alone (MDACC: 24% vs 7% and Moffitt Cancer Center: 17% vs 7%).[4, 5] In both studies, the use of adjuvant XRT was also found to be associated with improved LC on multivariate analysis (MDACC: P = .009 and Moffitt Cancer Center: P < .001). The data from each group support the combination of high-quality wide-margin surgery with adjuvant XRT as the optimal approach to local therapy. In addition, the size of these studies and the authors' detailed analyses help us to determine which subsets of the population are the most appropriate candidates for XRT.

The data indicate an absolute requirement for adjuvant XRT in patients with a positive surgical margin. Both groups identified statistically significant differences when comparing LC among patients with positive and negative surgical margins, independent of the treatment delivered. Strom et al reported a local recurrence rate of 29% versus 10% (log-rank P < .001) in patients with positive compared with negative surgical margins[5] and Guadagnolo et al demonstrated 10-year LC rates of 86% versus 49% (P = .001),[4] also favoring those patients with negative surgical margins. In the study by Strom et al, the effect of adding XRT was assessed for the 35 patients with positive surgical margins. There was a dramatic difference in local failure (LF) for those patients treated with surgery alone (7 of 13 LFs; 54%) compared with those receiving adjuvant XRT (3 of 22 LFs; 14%) (P = .003). There was a relatively small number of patients with positive surgical margins in the MDACC group,[4] and therefore a similar comparison could not be made. These data confirm the findings of multiple earlier studies that the presence of a positive surgical margin leads to increased LF and is therefore an indication for adjuvant XRT.[3, 11-13]

The recommendation for adjuvant XRT is more nuanced for patients with pathologically negative surgical margins. Strom et al demonstrated a trend toward improvement in all patients with a negative surgical margin who received XRT (6% vs 13%; P = .09)[5] and Guadagnolo et al demonstrated a difference in their entire patient population, which consisted primarily of patients with negative surgical margins (only 7% of patients had positive surgical margins).[4] Within this subpopulation, the key question is whether we can distinguish between patients who need and those who do not need postoperative adjuvant XRT. On univariate analysis, Strom et al demonstrated statistically significant improvements with the addition of XRT for those patients with negative surgical margins and any 1 of the following 3 features: head and neck location of the tumor, Breslow depth > 4 mm, and Clark level V tumors.[5] These characteristics have previously been shown to be associated with poorer local control in patients with DM,[1, 3, 9, 13, 14] and therefore the improvements noted with adjuvant XRT are not unexpected.

The fourth feature in this category is the malignant involvement of neural structures (neurotropism). Neurotropism is a rare finding if one considers all cutaneous melanomas, but the entity of desmoplastic neurotropic melanoma (DNM) can represent as much as 78% of the patients in a DM series.[7, 15] Both of the studies published herein[4, 5] presented evidence suggesting a benefit from local XRT in the setting of neurotropism. Strom et al noted that the patients with nerve involvement “might” benefit from XRT based on the Kaplan-Meier curves and a 5-year actuarial LC difference of 18% with the addition of XRT.[5] In the study by Guadagnolo et al, the presence or absence of perineural invasion (PNI) did not lead to a difference in LC among the entire population.[4] However, within the group of patients with evidence of PNI, the addition of XRT led to an improvement in LC at 10 years of 91% versus 63% (P = .02).

Thus, for patients with negative surgical margins and any of the 4 features discussed earlier, adjuvant XRT should be used. One prior single-institution study did demonstrate a very low rate (4%) of local disease recurrence in patients treated with wide local excision alone,[16] raising questions about the factors discussed above. The extent of surgery was very tightly controlled in the study: for lesions measuring > 2 mm in depth, the margins were 2 cm; if the lesions were 1 to 2 mm in depth, the margins were 1 to 2 cm; and if lesions measured < 1 mm in depth, the margins were 1 cm.[14] For the current studies, Strom et al stated that the “majority” of their patients had margins of > 2 cm; in regions with complex anatomy, margins were at least 1 cm.[5] Guadagnolo et al indicated that approximately 94% of their patients underwent “wide local excisions.”[4] It should be noted that this degree of success with surgery alone[16] has not been reported in other studies.

In addition to the differences between DM and other cutaneous melanomas in terms of local disease control, DM also has a distinct behavior in terms of regional lymph node involvement.[7] The standard initial evaluation of all patients with cutaneous melanoma includes an SLNB. The technique is accurate for assessing regional disease spread and the presence of malignant lymph nodes is the critical factor in determining survival. Although this is standard practice for melanomas overall, some data have suggested that the rate of lymph node disease in patients with DM is so low that George et al[17] have proposed skipping the SLNB. However, this difference in patterns of lymph node spread may only become apparent if tumors are classified into pure and mixed DM.[17, 18] The data presented by Strom et al do not support this approach, because they reported that approximately 12% of their patients with “pure” DM had positive SLNB findings.[5] Consequently, the current consensus remains to perform the SLNB procedure until definitive evidence emerges.

The data from the 2 publications regarding regional lymphatic treatment are less clear than for the primary tumor site. Guadagnolo et al reported that of the 71 patients treated with XRT, 25% received radiation to the draining lymphatics.[4] At the time of presentation, only 4% of patients were found to have lymph node involvement. Even with this small percentage, their data indicated that lymph node involvement is a very poor prognostic sign. Of the 130 patients in the study, 64 were classified with clinical N0 disease at the time of presentation and 56 underwent SLNBs, all of which were found to be negative for lymph node disease.[4] The actuarial rates of local recurrence were 11% at 5 years and 14% at 10 years, although no comments were made regarding whether there was treatment of the lymph node basins with respect to subsequent failure. Strom et al discussed their criteria for delivering XRT to the lymph nodes based on the presence of the following features: lymph node size > 3 cm, extracapsular extension, or malignancy in ≥ 3 lymph nodes. They stated that 6 patients received XRT to the lymph nodes after lymphadenectomy.[5] They reported a regional recurrence rate of 8% and did note improvement in combined locoregional control if XRT was added.[5] Factors identified as being associated with poor locoregional control were age > 70 years and a lesion depth > 4 mm. XRT to the lymph nodes was used infrequently because of rare lymph node spread, and therefore Strom et al believed they could not comment on this component of treatment.[5] Thus, the recommendation would be to consider regional lymphatic treatment if any high-risk factors are noted, but to not use it routinely.

The final question regarding the extent of the radiation fields relates to the nerve pathways. Do they need to be covered if neurotropism is identified? As noted above, this is an important question due to the high percentage of DM lesions classified as desmoplastic neurotropic melanoma. Guadagnolo et al state that PNI was assessed in 101 patients and found to be present in 66 patients.[4] They describe only 1 potential failure in a nonradiated patient that occurred in the setting of pathologic PNI. These 2 articles do not offer much data addressing this question; there is no strong indication to support routine comprehensive XRT coverage of nerve pathways. Thus, unless there is macroscopic nerve involvement, treatment of the nerve pathways should not be part of standard therapy.

Once the decision has been made to proceed with XRT and the appropriate regions designated for inclusion, what can the authors tell us about how this treatment should be delivered? Both groups used wide margins around the surgical site for the radiation fields (3-4 cm in the study by Guadagnolo et al[4] and 2-4 cm in the study by Strom et al[5]) and this is a requirement for adjuvant therapy for patients with DM. It is important to note that these wide margins will include a portion of the nerve pathways, even if they are not explicitly targeted. There were a variety of dose fractionation combinations used, ranging from the melanoma-specific hypofractionated regimen of 6 grays (Gy) per fraction to a total dose of 30 Gy to essentially standard fractionations of 1.8 Gy to 2.0 Gy to 60 Gy. No differences in outcomes were identified between these regimens. Thus, these and other intermediate-dose fractionation combinations can be considered,[19] although the current bias is to use higher doses per fraction than standard fractionated XRT. The hypofractionated schedule can be delivered safely, but it must be done with great care due to the high dose per fraction. Radiation oncologists should note the technical point emphasized by both groups that for hypofractionated treatment with electrons, dose is prescribed to the maximum dose to minimize potential complications.[4, 5]

In making the decision to add XRT, we need to consider the potential clinical costs for the patient. Guadagnolo et al were able to more comprehensively assess treatment-related toxicity because all XRT was delivered at their 1 center. Nine of the 15 reported toxicities were mild to moderate and were consistent with the morbidity of XRT to the head and neck including fibrosis, altered wound healing, and hypothyroidism. Six patients reported more severe issues such as osteoradionecrosis and surgical issues including graft failure.[4] These complications are severe, although rare, and can typically be managed. These risks must be weighed against the implications of a local disease recurrence and these considerations favor the pursuit of adjuvant therapy.

In summary, the published data support a role for adjuvant XRT to the primary site of a resected DM in all patients with a positive surgical margin. For patients with negative surgical margins and one of the identified risk factors, XRT should be delivered. The regional lymphatics should not be treated routinely, but should be considered for inclusion only if an additional risk factor is present. The treatment of nerve pathways can be considered in exceptional cases. In the future, these recommendations can hopefully be refined with the incorporation of the current pathologic classification system and additional clinical data such as those presented by Guadagnolo et al[4] and Strom et al.[5]

FUNDING SUPPORT

  1. Top of page
  2. FUNDING SUPPORT
  3. CONFLICT OF INTEREST DISCLOSURES
  4. REFERENCES

No specific funding was disclosed.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. FUNDING SUPPORT
  3. CONFLICT OF INTEREST DISCLOSURES
  4. REFERENCES

The author made no disclosures.

REFERENCES

  1. Top of page
  2. FUNDING SUPPORT
  3. CONFLICT OF INTEREST DISCLOSURES
  4. REFERENCES