Abstract 855P
Background
While immune checkpoint inhibitors (ICI) have revolutionized treatment of metastatic melanoma (MM) patients, still 40–60% of patients do not achieve a clinical benefit. Tissue-based predictive biomarkers are not validated in MM, and non-invasive liquid biomarkers can be an alternative source. Whole-blood transcriptome has shown to identify early response to ICI in urothelial cancer, thus have a potential in the context of MM.
Methods
Whole blood samples from 29 patients with BRAF+ and high lactate hydrogenase MM were collected before and after 6 weeks of anti-PD1/CTLA4 treatment. Nineteen were classified as responders (R) (progression-free survival (PFS) ≥ 6 months) and 10 as non-responders (NR). Transcriptome profiles were generated by RNAseq and differentially expressed genes (DEG) identified by combining differential expression and multivariate analysis. Biological relevancy was assessed by over representation and gene network analysis and performance was evaluated by Sparse Partial Least Squares (SPLS) regression and cross validation.
Results
One hundred nineteen DEGs have been identified to predict ICI response at baseline (BL). The list is enriched for genes related to CD8 T-cells as well as interferon and TLR signaling, suggesting R and NR pts have different immune cell composition at BL. Seventy and 400 DEGs have been identified in NR and R, respectively, by comparing BL and on-treatment samples. Although cell proliferation genes were commonly identified in R and NR, suggesting ICI can exert an effect in both groups, genes associated to cell chemotaxis, anti-PD1/CTLA4 therapy targets and interferon-γ were found specifically associated to R, suggesting them as key processes for a durable response. SPLS model coupled with PFS analysis showed a highly significant difference at 6-month between the 2 patient strata (100% R vs 15% NR, p<0.0001).
Conclusions
Whole blood transcriptome profiling is a promising tool to identify predictive biomarkers of response to ICI in MM and to understand underlying biological processes. Future plans include biomarker validation on an independent sample cohort (under collection) to confirm performances and identified biological processes.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
E. Durandau: Financial Interests, Personal, Full or part-time Employment: Novigenix SA. L. Ciarloni: Financial Interests, Personal, Full or part-time Employment: Novigenix SA; Financial Interests, Personal, Stocks/Shares: Novigenix SA. N. Hadadi: Financial Interests, Personal, Full or part-time Employment: Novigenix SA. S. Hosseinian Ehrensberger: Financial Interests, Personal, Full or part-time Employment: Novigenix SA; Financial Interests, Personal, Stocks/Shares: Novigenix SA; Financial Interests, Personal, Member of the Board of Directors: Novigenix SA. N. Mehra: Financial Interests, Personal, Advisory Board: Pfizer, Roche, MSD, AstraZeneca, Astellas, JNJ; Financial Interests, Institutional, Advisory Board: Janssen; Financial Interests, Institutional, Funding: Astellas, Pfizer; Financial Interests, Personal and Institutional, Funding: Janssen; Financial Interests, Institutional, Invited Speaker: BMS, Janssen, BMS; Financial Interests, Institutional, Research Grant: AstraZeneca, BMS; Non-Financial Interests, Leadership Role, Head of the Prostate Cancer Working group: Dutch Uro-Oncology Study Group; Non-Financial Interests, Principal Investigator, co-PI: Prospective Bladder Cancer Infrastructure (Netherlands); Non-Financial Interests, Leadership Role: Castration-resistant Prostate Cancer Registry. All other authors have declared no conflicts of interest.