Jeffrey M. Conroy1, 2, Mary Nesline1, Sarabjot Pabla1, Sean T. Glenn1, 3, Antonios Papanicolau-Sengos1, Blake Burgher1, Jonathan Andreas1, Vincent Giamo1, Moachun Qin1, Felicia L. Lenzo1,Devin Dressman1, Marc Ernstoff4, Igor Puzanov4, Mark Gardner1, Carl Morrison1, 2,*

1. OmniSeq Inc., 700 Ellicott Street, Buffalo, NY 14203, US, 2. Center for Personalized Medicine, Roswell Park Comprehensive Cancer Center, Elm and Carltons Streets, Buffalo, NY 14263, US, 3. Cancer Genetics and Genomics, Roswell Park Comprehensive Cancer Center, Elm and Carltons Streets, Buffalo, NY 14263, US, 4. Department of Medicine, Roswell Park Comprehensive Cancer Center, Elm and Carltons Streets, Buffalo, NY 14263, US

* Dr. Carl Morrison, MD, DVM: Carl.Morrison@OmniSeq.com

ASCO-SITC 2018

Comprehensive Immune and Mutational Profile of Melanoma

Copyright 2018 OmniSeq, Inc.

Conclusions

Figure 5: The mutational status of melanoma impacts the tumor immunemicroenvironment. Triple WT tumors (top panel) are over represented in all immunephenotypes while driver mutation tumors (panels 2 – 4) are under represented. CDKN2Aloss (data not shown) was also under represented in majority of immune phenotypes.

References1. Conroy JM, Pabla S, Glenn ST, et al. Analytical Validation of a Next-Generation Sequencing Assay to Monitor Immune Responses in Solid Tumors. J Mol Diagn 2018 Jan;20(1):95-109.2 Akbani R, Akdemir KC, Aksoy BA, et al. Genomic Classification of Cutaneous Melanoma. Cell 2015;161(7):1681–96.

Statistical analysis (v.test) of the four genomic subtypes (RAS mutant, BRAFmutant, NF1 mutant and triple WT) demonstrated over-representation of allimmune phenotypes for the Triple WT cohort, and under-representation ofimmune phenotypes for subtypes harboring a driver mutation (Figure 5).

BRAF/RAS/NF1 mutant subtypes are immunophenotypically distinct fromTriple WT and do not associate with an inflamed tumor microenvironment,suggesting that these most common melanoma driver mutations areassociated with “cold” tumors that may not promote an adaptive anti-tumor response. Triple WT samples present with an overall activatedimmune phenotype, representative of an inflamed “hot” tumor. Thesefindings suggest that an integrated genomic and immunophenotypicapproach is necessary to better understand the immune microenvironment,the effect of genomic alterations on immune response and implications toimmunotherapy modalities.

Results

Methods

Introduction

306 metastatic melanoma samples were tested by NGS using acomprehensive cancer panel for mutational status and an immune responsepanel which interrogates the expression profile of 54 validated immune-related genes1 (Figure 1).

The introduction of checkpoint inhibitor immunotherapy, predominantlyanti-PD-1 and anti-CTLA-4 monoclonal antibodies, has provided significantclinical benefit for melanoma patients. The tumor microenvironment (TME)plays an essential role in therapeutic response and also tumor progression.The development of predictive biomarkers that measure overall tumormutational burden, microsatellite instability and PD-L1 expression arewidely used to characterize the immune TME and guide therapy selection.However, the genomic determinates that trigger an immunologically activetumor are not clear. In this study, we examined somatic mutations inmetastatic melanoma samples to determine if there is an associationbetween tumor mutational profiles and immune phenotypes, as measuredby next-generation sequencing (NGS). The findings indicate that themutational landscape effects the immune composition of the TME.

Figure 1: NGS workflow

Figure 2: Immune phenotypes

Tumors evaluated had histologically confirmed, previously untreated stageIII or IV melanoma and the availability of tumor tissue from a metastatic orunresectable site for NGS assessment of somatic mutations and geneexpression of immune markers. The level of gene expression was rankedagainst a reference population and represented as seven immunephenotypes (Figure 2).

Somatic mutations were classified into four genomic subtypes: triple wild-type (WT), BRAF mutant, RAS mutant (NRAS, HRAS or KRAS), and NF1mutant, as identified by The Cancer Genome Atlas (TCGA)2. Copy numbergain and loss were limited to focal amplifications >4 copies, andhomozygote loss (< 1 copy), respectively.

Results

Figure 3: Total number of somatic mutations, frequency of mutation subtypes (Triple WT,BRAF, RAS and NF1) and other variants detected in 306 metastatic melanomas.

Figure 4: The first principal component highly correlated with BRAF mutations (p < 0.001), thesecond highly correlated with RAS mutations (p < 0.001), and the third principal component,although not informative, highly correlated with NF1 mutations (p < 0.001).

Principal component analysis (PCA) was performed to determine association ofBRAF/RAS/NF1 mutations and Triple WT with immune phenotypes and immuneresponse gene expression as measured by the NGS panels. PCA demonstratedthat the first and second dimension explain 86% of the variation in the mutationprofiles of the 306 melanomas.

-3 -2 -1 0 1 2 3Over-representation (v.test)

Immune Phenotypes

T cell Primed Pro-inflammatory Myeloid Suppression Immune Escape Checkpoint Blockade Other Checkpoint Blockade (PD-L1/CTLA4) Anti-inflammatory

BRAFTriple WT

NF1

RAS

Trip

le W

TR

AS

NF1

BR

AF

38% cases were positive for activating BRAF mutations, 16% for RAS, and 7% forNF1. 46% were considered triple wild type and 51% harbored a CDKN2A (p16)loss of function (Figure 3).

700 Ellicott Street | Buffalo NY, 14203The right drug or right trial…

For Every Patient