Effect of Radiotherapy Dose and Volume on Relapse in Merkel Cell Cancer of the Skin
IJROBP Volume 77, Issue 3, Pages 677-684 (1 July 2010)
To assess the effect of radiotherapy (RT) dose and volume on relapse patterns in patients with Stage I–III Merkel cell carcinoma (MCC).
This was a retrospective analysis of 112 patients diagnosed with MCC between January 2000 and December 2005 and treated with curative-intent RT.
Of the 112 evaluable patients, 88% had RT to the site of primary disease for gross (11%) or subclinical (78%) disease. Eighty-nine percent of patients had RT to the regional lymph nodes; in most cases (71%) this was for subclinical disease in the adjuvant or elective setting, whereas 21 patients (19%) were treated with RT to gross nodal disease.
With a median follow-up of 3.7 years, the 2-year and 5-year overall survival rates were 72% and 53%, respectively, and the 2-year locoregional control rate was 75%. The in-field relapse rate was 3% for primary disease, and relapse was significantly lower for patients receiving ≥50Gy (hazard ratio [HR] = 0.22). Surgical margins did not affect the local relapse rate. The in-field relapse rate was 11% for RT to the nodes, with dose being significant for nodal gross disease (HR = 0.24). Patients who did not receive elective nodal RT had a much higher rate of nodal relapse compared with those who did (HR = 6.03).
Conclusion
This study indicates a dose-response for subclinical and gross MCC. Doses of ≥50Gy for subclinical disease and ≥55Gy for gross disease should be considered. The draining nodal basin should be treated in all patients.
Discussion
The optimum treatment for MCC of the skin remains controversial because there is a lack of data on which to base treatment algorithms. With the mounting evidence that RT improves locoregional control and survival, it is important that the appropriate radiation doses and volumes are defined in the curative management of these patients. This cohort of 112 patients reflects the largest series attempting to define these RT parameters.
The use of retrospective data to estimate dose-response has its limitations. Patients are typically not distributed evenly over the range of doses; approximately one third of all doses in this study were 50 Gy. If the low and high ends of the dose scale are given to specific patients then there is potential for confounding bias. Dose responses in this study were not affected by adjustments for other variables, suggesting that confounding, if any, is minimal.
Generally, the current literature suggests that the optimum treatment for MCC is with a combination of surgery plus RT in terms of recurrence and survival. Details of trials evaluating the role of adjuvant RT published within the last 10 years (for >30 patients) are outlined. It can be seen that the addition of locoregional RT to surgery decreases local and regional recurrence rates.
In the largest series documenting the natural history of this malignancy a wide local excision was attempted on almost all patients with a known primary, with a 94% rate of negative margins. In our series 19% of patients had either positive histologic margins or gross local disease. Despite the less-extensive local surgery in this cohort, the rate of local recurrence was only 12%, with an in-field local recurrence rate of 3%. On univariate analysis gross disease or an excision with positive margins had no influence on the rate of local relapse either in- or out-of-field. This rate of local recurrence is comparable to series in which more extensive surgery was undertaken and suggests that with the addition of RT surgical margin status is not critical.
The meta-analysis by Gupta showed that for patients with clinical Stage I and II disease the rate of pathologic node involvement by sentinel lymph node biopsy (SLNB) was 32%. This is similar to our rate of 33% for nodal relapse in patients with Stage I and II disease who did not receive nodal surgery or RT. This nodal relapse rate is significantly higher than the rate of 8% for those who had elective nodal RT for early-stage disease (multivariate HR = 6.03). Because of the high rate of pathologically positive nodes in early-stage (I/II) disease and the effectiveness of elective nodal RT, together with the potential morbidity associated with staging the nodal basin, we recommend elective nodal treatment in all early-stage (I/II) patients in the absence of negative results on SLNB.
To analyze different dose and fractionation schedules in this cohort all doses were converted to 2-Gy equivalents according to the Withers formula, with an α/β of 10. It is noteworthy that patients who received hypofractionated treatment had doses that were somewhat lower using this adjustment, and the reduction in overall treatment time may be significant in a rapidly proliferating tumor. To establish a dose-response, we classified both local and nodal failures as in-field or out-of-field. We believe that this is justified given that out-of-field failures are a reflection of inadequate RT volumes as opposed to dose. suggests that there were no in-field failures in the primary or nodal sites for subclinical disease for doses of >50 Gy.
It is of interest that dose ≥50 Gy was a significant factor for local relapse for local gross and subclinical disease (HR = 0.28). In addition, for subclinical disease, there were no in-field relapses for doses >50 Gy, and out-of-field relapse was also low for doses >50 Gy. The reason for the lower rate of out-of-field relapse in relation to radiation dose is difficult to understand. Patients who had hypofractionated treatment had relatively lower doses than standard fractionation and are overrepresented in the group that had out-of-field relapse.
For nodal RT there is a dose-response for gross disease, and as outlined there seems to be a dose-response for subclinical disease. As displayed, there seems to be little benefit for doses >50 Gy for subclinical nodal disease and >55 Gy for gross nodal disease. We have shown that a dose of 50 Gy seems adequate for elective and adjuvant treatments. Higher doses are required for gross disease and should be tailored to the burden of disease.
There is increasing interest regarding the role of chemotherapy in MCC of the skin, particularly for those patients with high-risk disease. In this series, which includes a large proportion of patients with Stage I disease, chemotherapy was used in 15% of patients. As such there are insufficient numbers to make any comment regarding the addition of this in terms of locoregional and distant metastatic control.
On the basis of the data presented here, we can make recommendations of 50 Gy for subclinical disease and >55 Gy for gross disease. In most patients with local relapse this was out-of-field, generally in close proximity to the field edge. Almost half of the patients with regional relapse had out-of-field relapse. To reduce the rate of out-of-field relapse, we recommend increasing the field size for local RT to encompass the primary disease with a minimum 4-cm radial margin where practical. Although hypofractionation may be reasonable in some patients, this should not be at the expense of a reduction in field size. For elective, adjuvant, and radical nodal RT we recommend treating the entire draining lymph node basin where possible.
Conclusions
The important role of RT in reducing local and nodal relapses has been confirmed. There seems to be a dose-response up to 50 Gy for subclinical disease and for gross disease at >55 Gy. These doses achieve very high in-field local control rates. The problem of out-of-field failures can only be lowered by more generous field sizes. Obtaining wide surgical margins does not seem to offer improved control rates, providing RT is to be incorporated into the treatment regimen. For early-stage (I/II) disease we recommend elective nodal RT in the absence of negative results on SLNB.