The interferon-gamma (IFN-y) signature from baseline tumor material predicts pathologic response after neoadjuvant ipilimumab (IPI) plus nivolumab (NIVO) in stage III melanoma.

9539 Background: Neoadjuvant IPI + NIVO induces high pathologic response rates (pRR) of 74-78% in macroscopic stage III melanoma. Pathologic response ( < 50% viable tumor) is strongly associated with improved relapse-free survival (RFS); the previous OpACIN-neo study demonstrated a 2-year RFS of 96.9% in patients (pts) with pathologic response, whereas the 2-year RFS in non-responders was 35.5%. These data highlight the need for baseline biomarkers predictive for response and survival. Here, we present the predictive value of the 10-gene IFN-y expression signature algorithm (based on Ayers et al.) for pathologic response and relapse in a cohort of melanoma pts treated with neoadjuvant IPI + NIVO. Methods: Baseline tumor biopsies from lymph node metastases of stage III melanoma pts were used for IFN-y signature assessment. Pts were treated with a maximum of two cycles of neoadjuvant IPI 1mg/kg + NIVO 3mg/kg in the OpACIN-neo (arm B) and PRADO studies. RNA expression analysis was conducted using the nCounter® PanCancer Immune Profiling panel on the NanoString Flex machine (NanoString Technologies), which is clinically applicable due to its fast turn-around-time (two days). An IFN-y signature gene expression score (IFN-y score) was calculated using a NKI-developed algorithm. Association between IFN-y score and pathologic response or event-free survival (EFS) was examined by logistic regression and Cox analyses. The optimal cutoff between a high and low IFN-y score was defined based on a summary receiver operating characteristic (sROC) curve. Results: In total, 103 pts treated in the OpACIN-neo and PRADO studies had baseline tumor material available. Median age was 56 years, 62% was male, and 52% had a high baseline IFN-y score. The pRR of the total cohort was 70% (72/103 pts), including 56% (58/103) major pathologic response (MPR, 0-≤10% viable tumor) and 14% (14/103) partial responses (pPR, 10-≤50% viable tumor). 30% (31/103 pts) had no pathologic response. After a median follow-up of 25.2 months, 26 pts (25.2%) developed a melanoma relapse. The IFN-y score was significantly associated with response (OR 1.061, p < 0.001) and relapse (OR 0. 974, p = 0.029). The pRR was 89% (48/54) in pts with a high IFN-y score versus 49% (24/49) in those with a low IFN-y score (p < 0.001). Pts with a high IFN-y score were also less likely to develop a relapse (11% [6/54] versus 41% [20/49], p = 0.001). Conclusions: Pts with a high IFN-y score in pre-treatment biopsies are more likely to respond to neoadjuvant IPI + NIVO with favorable EFS. A rapid gene expression analysis enables the IFN-y score to be used in daily clinical practice to identify pts who might qualify for treatment escalation or de-escalation. The DONIMI study [NCT04133948] currently investigates different neoadjuvant treatment combinations in stage III melanoma pts based on their intratumoral IFN-y score.