Plasma and exosome proteomic profiling for prediction of immunotherapy response and toxicity

Arnav Mehta1,2, Gyulnara Kasumova1, Alvin Shi3, Marijana Rucevic4, Markus Sallman-Almen4, Lina Hultin Rosenberg4, Emmett Sprecher4, Jacqueline Ohmura1, Michelle Kim1, David Lieb2, Xue Bai1, Dennie T. Frederick1, Russell Jenkins1, Manolis Kellis3, Ryan J. Sullivan1, Keith T Flaherty1, Nir Hacohen1,2, Genevieve M. Boland1 1Masshusetts General Hospital, Boston, MA, 2Broad Institute of MIT and Harvard, Cambridge, MA, 3MIT, Cambridge, MA, 4Olink Proteomics, Watertown, MA

Background Correlates of anti-PD1 treatment Exosome proteomic markers differentiate R vs NR

The response of metastatic melanoma to anti-PD1 is 70 significantly differentially expressed (DE) were Analysis of exosome proteomic markers revealed trends heterogeneous. Several non-invasive biomarkers are identified across the treatment period, including markers to suggest differentially expressed immune proteins. currently under investigation as diagnostic strategies of immune activation (PD1, CXCL9, CXCL10, CD25, IL-17a, These proteins include CXCL9, CXCL10, CXCL11, CXCL13 including circulating tumor cells (CTCs), cell free DNA or among others). and CCL9, among others.

RNA, and exosomes. To date, the best predictors of CXCL9 CXCL10 CXCL11 CXCL13 IFNg PDL1 PDL2 response involve tissue biopsy for immune profiling. It is likely the most effective non-invasive ~1,000 plasma biomarkers predictors will involve layered datasets, utilizing several source of information. Towards this effort, we performedin 55 Melanoma patients proteomic profiling of patient plasma and exosome samples to build a predictor of immunotherapy response and uncover biological insights underlying primary Comparison of plasma proteins to tumor scRNA-seq resistance. Methods Analysis of single-cell RNA-seq data (Sade-Feldman et al, 58 metastatic Cell, 2018) of tumor tissue from a subset of these melanoma patients patients revealed that expression of most proteins predictive of response were enriched among tumor receiving anti-PD1 at 8 5 Moshe_2017−10−16_Melanoma_Keren Plasma sampling olink4 Serum proteomic markers differentiate anti-PD1 1 myeloid cells, with the remainder of proteins being MGH comprised the Pre-treatment initial cohort, and 150 responders (R) and non-responders (NR) reflective of exhausted T cell states. On-treatment (Histogram) 0 1 2 3 4 5 RESPONDERS Post-treatment 14 additional patients cluster B-cells 38 significantly DE proteins identified with on-treatment IL12RB1 PDCD1 CXCL13 Plasma cells TNFRSF9 (currently being VCAM1 HAVCR2 Monocytes/macrophages time points between29 anti-PD1 responders and non- NON-RESPONDERS CD27 TNFRSF1B CRTAM Dendritic cells KLRD1 processed) comprised LAG3 HAVCR1 SELL Lymphocytes responders, including several implicated in primary or LY9 EPHA2 CLEC4C FLT3LG Exhausted CD8+ T cells a validation cohort. CD28 CD6 CD5 Regulatory T cells acquired resistance (IL8, MIA, among others). NOS3 IL17A ERBB2 Cytotoxicity (lymphocytes) IL1RL1 Plasma samples were LAIR2 IL2RA Exhausted/HS CD8+ T cells st CD177 SPINK1 SDC1 Anti-PD-1 Nivolumab (BMS) 1 line: Anti-CTLA-4 Ipilimumab (BMS) PILRB Memory T cells collected (MGH IRB CD1c CXCL13 CLEC5a ANGPT2 Responders SPARCL1 TNFRSF11A AREG Lymphocytes exhausted/cell-cycle nd CD1C CD274 Anti-PD-1 Pembrolizumab (Merck) 2 Nonline:-responders Anti-PD-1 Nivolumab (BMS) TNFSF10 #11-181) at baseline ● ● ● ● CXCL9 ● CXCL10 CXCL11 TNFRSF6B IL10 ANGPTL4 STXBP3 and several on- QDPR Days after start of treatment ● SIRPB1 LILRB4 ● ● CST3 TNFSF13 ● ● THBD treatment time-points. (average 67 days from baseline) ● ● LILRA5 LGALS9 IL18 SLAMF8 DSC2 LAIR1 SIGLEC1 CCL2 CLEC6A SIGLEC10 QPCT CD300LF CLEC7A IL8 IGF1R AREG MIA PILRA Samples were analyzed for 1102 proteins by a multiplex TREM1 VCAN OSCAR CD163 VSIG4 Proteomic profiling of biomarkers for response to immunotherapy in melanoma FBP1 proximity extension assay (Olink Proteomics). A subset of ● ● CLEC5A

● ● 1 2 3 4 5 6 7 8 9 ● 10 11 patients using Proximity Extension Assay ● patients had single-cell RNA-seq (Smart-Seq2 protocol) ● ● ● ● ● ● Conclusions ● performed on tumor tissue. Group differences and ● ● ● IMMUNO REACTION EXTENSION REACTION AMPLIFICATION treatment effects were evaluated using linear mixed 2019 ASCO-SITC Immuno-Oncology Symposium Whole plasma and exosome proteomic profiling of anti- models with maximum likelihood estimation for model PD1 treated patients revealed DE proteins between R and Proteomic profiling of biomarkers for response to immunotherapy in melanoma Proteomic profiling of biomarkers for response to immunotherapy in melanoma Proteomic profiling of biomarkers for response to immunotherapy in melanoma parameters, and Benjamini and Hochberg multiple NR that may enable a liquid biopsy to predict anti-PD1 patients using Proximity Extension Assayhypothesispatients using Proximity Extension Assaypatients using Proximity Extension Assaycorrection. response. These results unveil a putative role of myeloid cells within the tumor microenvironment in anti-PD1 IMMUNO REACTIONIMMUNO REACTIONProximity extension assay (PEA)*EXTENSION REACTIONEXTENSION REACTION AMPLIFICATIONAMPLIFICATION DETECTION QC & DATA ANALYSIS IMMUNO REACTION EXTENSION REACTION AMPLIFICATION response or primary resistance. Acknowledgements

Protein detection with Extension reaction Amplification Detection and analysis We are thankful to the MGH Jackson Society White Coat DETECTIONDETECTIONpairs of antibodies QC & DATA ANALYSIS DETECTION QC & DATA ANALYSISQC & DATA ANALYSIS Grant for funding, and the rest of the Boland and *Olink Proteomics Hacohen labs for their support and discussions.