Prognostic value of autoantibodies (auto-AB) in melanoma patients (pts) in the EORTC 18952 trial of adjuvant interferon (IFN) compared to observation (obs)

8507 Background: Appearance of auto-AB during IFN-adjuvant therapy is a prognostic factor in melanoma pts. We evaluated this in the EORTC 18952 trial including pts without IFN. Methods: Anti-cardiolipin (C) anti-thyreoglobulin (T) and anti- nuclear (N) AB levels of serial samples between 0 and 48 months (mt) were assessed centrally. Prognostic impact of CTN on relapse-free survival (RFS) was assessed by 3 methods: 1) usual Cox model (lead-time bias); 2) Cox Time dependent model using the latest positive CTN value; 3) Cox Time dependent model using the latest any CTN value. Results: In 278 pts the presence of auto-AB at any time was determined. Among them, 187 pts (43 stage IIb, 144 stage III), Obs (32) or 13-mt-IFN (80) or 25-mt-IFN (75), had their initial auto- AB status assessed: 60 (32%) were CTN-positive (pos) (similar % in 3 arms) and 127 (68%) were CTN-negative (neg). Of these 127 pts, during the study, at least once, the following became CTN-pos: Obs (7/24=29%), 13-mt-IFN (19/55=36%), 25-mt-IFN (20/48=42%); 1-yr conversion rate was 23%, 34% and 38% in the 3 arms respectively; 2-yr rate was 23%, 46% and 61%. Median follow-up = 4.2 yrs. Relapse occurred in 122/187 pts. Median RFS = 2.1 years. In the 187 pts with a known initial CTN-status, using the classical Cox model the hazard ratio (HR) of those who were/became CTN-pos vs the others, was 0.50. In order to overcome the Lead Time Bias, the Cox time dependent model was used: the HR for pts with at least one pos CTN-status vs always CTN-neg was 0.63 and for those with the latest CTN-status pos vs those with latest CTN-status neg was 0.87. All analyses were adjusted for # of pos nodes. Results were similar when only IFN pts or Obs only pts were considered (test for interaction: P>0.50), or when all pts (278) or only pts CTN-neg at randomization (127) were assessed. Initial CTN status known. Conclusions: We could not confirm presence or appearance of auto-AB as a strong prognostic factor in melanoma pts and did not find a strong interaction with IFN therapy. [Table: see text] [Table: see text]