Background: Appearance of autoantibodies and clinical manifestations of autoimmunity in melanoma patients treated with adjuvant interferon (IFN)-alpha2b was reported to be associated with improved prognosis. We assessed the association of the appearance of autoantibodies after initiation of treatment with recurrence-free interval in two randomized trials that compared intermediate doses of IFN with observation for the treatment of melanoma patients.
Methods: Serum levels of anticardiolipin, antithyroglobulin, and antinuclear antibodies were determined using enzyme-linked immunosorbent assays in 187 and 356 patients in the European Organization for Research and Treatment of Cancer (EORTC) 18952 and Nordic IFN trials, respectively, immediately before and up to 3 years after random assignment. The association of the presence of at least one of the three autoantibodies with risk of recurrence was assessed by three Cox models in patients negative for all three autoantibodies at baseline (125 from the EORTC 18952 trial and 230 from the Nordic IFN trial): 1) a model that considered appearance of autoantibodies as a time-independent variable, 2) one that considered a patient autoantibody positive once a positive test for an autoantibody was obtained, and 3) a model in which the status of the patient was defined by the most recent autoantibody test. All statistical tests were two-sided.
Results: When treated as a time-independent variable (model 1), appearance of autoantibodies was associated with improved relapse-free interval in both trials (EORTC 18952, hazard ratio [HR] = 0.41, 95% confidence interval [CI] = 0.25 to 0.68, P < .001; and Nordic IFN, HR = 0.51, 95% CI = 0.34 to 0.76, P < .001). However, on correction for guarantee-time bias, the association was weaker and not statistically significant (model 2: EORTC 18952, HR = 0.81, 95% CI = 0.46 to 1.40, P = .44; and Nordic IFN, HR = 0.85, 95% CI = 0.55 to 1.30, P = .45; model 3: EORTC 18952, HR = 1.05, 95% CI = 0.59 to 1.87, P = .88; and Nordic IFN, HR = 0.78, 95% CI = 0.49 to 1.24, P = .30).
Conclusions: In two randomized trials of IFN for the treatment of melanoma patients, appearance of autoantibodies was not strongly associated with improved relapse-free interval when correction was made for guarantee-time bias.