Prognostic scoring in patients with melanoma after adjuvant isolated limb perfusion

BACKGROUND: The clinical course in melanoma is variable. The aim of the present study was to assess adjuvant isolated limb perfusion (ILP) efficacy using a surrogate comparison of observed survival versus Cochran-predicted survival. MATERIALS AND METHODS: All patients in a single university hospital with primary, non-ulcerated limb melanoma who had undergone adjuvant ILP over 10 years (1986-1995) were studied. Clinical and pathological details including follow-up and survival were prospectively recorded in a national database. All patients were risk scored, as described by Cochran et al., to yield individual survival probability at the end of 3, 5 and 10 years and this was compared with observed survival at corresponding intervals. RESULTS: There were 85 patients who had adjuvant ILP for primary non-ulcerated limb melanoma. Of these, 14 deaths were observed (O) within the 10-year follow-up period. The Cochran score predicted (E) 20 deaths within 10 years (O/E ratio 0.7). The O/E ratios for deaths in the 0 to 3, 3 to 5, and 5 to 10 year intervals were 8/7.4, 5/6.0, and 1/6.5, respectively; prediction of late deaths tended to be overestimated. When patients were grouped by predicted 10-year mortality (<20%, 20-40%, >40%) the overestimation was found to occur mainly in the highest risk group: O/E ratios were 6/5.9, 6/8.4, and 2/5.6, respectively (P = 0.10, Hosmer-Lemeshow test). CONCLUSION: The observed and expected survival in patients receiving adjuvant ILP at the end of 3 and 5 years are comparable. The Cochran scoring system overestimated deaths during the 5 to 10 year interval. It is not clear whether this observation is a consequence of ILP efficacy or inaccuracy of the Cochran score.