Adjuvant radiotherapy for cutaneous melanoma: Comparing hypofractionation to conventional fractionation

Chang. U FLA. IJROBP 2006;66:1051

The use of adjuvant radiotherapy (RT) has been controversial because of the long-held belief that melanoma was radioresistant, based on in vitro studies demonstrating that melanoma cells possess a broad shoulder on the cell survival curve and thus have a large capacity for DNA repair. As a result, hypofractionated RT schedules have been developed to counteract this perceived radioresistance, producing excellent locoregional control rates of 85% and higher. Radiation Therapy Oncology Group (RTOG) Trial 83-05 was a prospective randomized study comparing hypofractionation to conventional fractionation. The results showed no difference in partial or complete response rates between the two schedules, and the overall response rates were 70%. This study both refutes the concept that melanoma is radioresistant and questions the need for hypofractionated treatment schedules. Unfortunately, the duration of response to treatment was not reported in this trial and, thus, the true radioresponsiveness to each fractionation schedule is unclear.

Adjuvant RT has been advocated in the presence of high-risk features that predict for locoregional failure. The purpose of this study is to report our experience with adjuvant RT to patients with a high risk of locoregional relapse using both fractionation schedules. Our initial treatment strategy employed conventional dose fractionation. In 1990, investigators at M.D. Anderson first published their preliminary results of a Phase II study using a hypofractionated RT schedule with 600 cGy fractions given twice per week on Monday and Thursday or Tuesday and Friday to a total dose of 3,000 cGy. The results were encouraging, with locoregional control rates exceeding 80%. Since the publication of that report, we have adopted this treatment schedule as standard treatment for cutaneous melanoma at the University of Florida.

Patients in this study were treated because of the following indications: gross disease, close or positive margins, disease recurrence, satellitosis, and regional node metastases. Patients who were treated electively to the regional nodes and were without high-risk pathologic factors were excluded. Only 1 patient was treated with gross disease. Forty-one patients (73%) were treated with the MD Anderson hypofractionated schedule administering five fractions of 600 cGy each twice per week (Monday and Thursday or Tuesday and Friday) over 2.5 weeks for a total dose of 3,000 cGy. If the spinal cord was in the initial field, it was excluded after 2,400 cGy. Fourteen patients (25%) were treated with once-daily fractionation to a median dose of 6,000 cGy (range, 5,000–7,000 cGy) at a median dose per fraction of 200 cGy (range, 171–200 cGy). One patient received twice-daily fractionation to 7,440 cGy at 120 cGy per fraction. Of the 49 patients with head-and-neck disease, 39 received hypofractionation (80%), whereas 2 of 7 patients with disease below the clavicles received hypofractionation (29%). Although distant metastases remain the predominant cause of mortality, control of locoregional disease is an important part of treatment. In the presence of adverse pathologic features, regional recurrence rates after surgery alone range from 30 to 50%. The results of surgery and adjuvant RT are superior to the historical control rates after surgery alone. Our locoregional control rate of 87% is comparable with other published data.

The historical concept that melanoma is radioresistant has led to the use of larger fraction sizes to overcome the high repair capacity of the tumor cells. Although good control rates are reported with the various hypofractionation schedules, no definitive evidence is available that suggests conventional fractionation is inferior. Corry  reported on 42 patients treated with 2-Gy fractions to 50–60 Gy and observed a local control rate of 80%, which is similar to those of reports using hypofractionation. The RTOG 83-05 trial showed no difference in response rates between 32 Gy in 4 fractions vs. 50 Gy in 25 fractions. However, this trial has been criticized for failing to provide data on the duration of response, a more accurate indicator of the effectiveness of treatment. Our data show no statistically significant difference in locoregional control between conventional fractionation and hypofractionation with a 5-year ifLRC of 87% in both arms (p = 0.966). Although these data must be viewed with caution because of the small patient numbers and possible selection biases inherent in retrospective studies, it does support the use of conventional fractionation, especially when hypofractionation is not preferred.

The treatment policy at our institution reflects these findings. There is concern of increased late effects with larger fraction sizes, which is consistent with our observation that the two severe late complications occurred in the hypofractionation group of patients. This risk is likely compounded if larger radiation field sizes are used. On the other hand, the hypofractionation schedule carries with it the advantage of patient convenience, requiring only five fractions given over 2.5 weeks. Thus for patients who have poor prognoses, are medically frail, or require only a modest treatment area (e.g., parotid with or without ipsilateral neck), hypofractionation would be appropriate. However, in situations in which the disease is adjacent to critical neurologic or optic structures sensitive to larger fraction sizes or that require large radiation fields, conventional fractionation is preferred to reduce the risk of late complications. An additional situation that may not be well-suited for hypofraction, and would likely be treated conventionally, is disease location on the scalp, because large fraction sizes could lead to a risk of bone exposure.

The only factor associated with infield locoregional relapse in our series was satellitosis. Other factors that have been associated with recurrence have been extracapsular extension, nodes >3 cm, positive margins, recurrent nodal disease, and more than three nodes involved. These factors were not associated with an increased risk of recurrence in our series, nor were the number of risk factors. Although the sample size is too small to make definitive conclusions, these encouraging findings may imply that in the face of adverse features, RT was an effective treatment modality to offset them. This interpretation is supported by Ballo et al., who found no disease characteristics associated with locoregional recurrence in their series of 466 patients treated with adjuvant RT at the MD Anderson Cancer Center (25). The only factor associated with FFDM, CSS, and OS was nodal involvement. This finding is also in accordance with the MD Anderson series, which showed inferior disease-specific survival and FFDM with increasing numbers of metastatic nodes.

Radiation therapy was well tolerated by patients. Acute toxicity was minimal with hypofractionation, with no patient requiring a treatment break. One patient was given antibiotics for acute parotitis that resolved soon after treatment. Otherwise, no other Grade 2 or any Grade 3 acute toxicity occurred. Serious late complications developed in 2 patients, both of whom received hypofractionation for a crude rate of 5%, for that subset of patients, and 4% for the overall group. This compares favorably with other reported series that used this fractionation schedule. Ballo et al. reported that the main late complication in the MD Anderson series was lymphedema, which was primarily associated with RT to the groin and axilla. Because our study had few patients with these sites involved, it is not surprising that we observed no cases of lymphedema. Another study from M.D. Anderson described 154 patients with 1 Grade 3 late complication of temporal lobe necrosis causing seizures and requiring surgery.