Stergios Moschos, MD Clinical Associate Professor of Medicine Department of Medicine Division of Hematology/Oncology University of North Carolina at Chapel Hill
Solid Tumor with one of the Highest Mutation Rates Most Frequent Genetic Aberrations are: BRAF V600, NRAS Q61, NF1, CDKN2A locus, PTEN, TP53 Largely Incurable, Even by 2014 Standards, with Short Overall Survival (OS, median 12 months) Chemotherapy-Resistant Host Immune Response Plays a Role The Role of Tumor Angiogenesis is Controversial 6 FDA-approved Therapies since 2011 Immunotherapies Pegylated Interferon Ipilimumab (CTLA4 blocking mab) Pembrolizumab (PD-1 blocking mab) Targeted Therapies Vemurafenib Dabrafenib Trametinib
Solid Tumor with the Highest Propensity for CNS Tropism Accounts for 50% of Mortalities from Brain Metastatic Melanoma High Propensity to Bleed The Most Frequent Oncogenic BRAF/NRAS Mutations Do Not Have a Prognostic Role in Established Brain Metastases Ipilimumab, Vemurafenib, and Dabrafenib Have a Definite but Short-Term Activity in Patients with Active Brain Metastases
High Immune Infiltrate (n=56) Low Immune Infiltrate (n=44) High Immume Infiltrate & Low Hem (n=19) All remaining (n=81) p=0.006 p<0.001 High Hemorrhage (n=58) Low Hemorrhage (n=42) p=0.040 Peritumoral CD8 + cells were prognostic FoxP3 + cells were not prognostic Hamilton Cancer 2013
Whole Genome Expression Profiling Revealed Biocarta Pathways Associated with OS Hamilton Cancer 2013
Is the Presence of, or Lack of Hypoxia, or Signaling Pathways within Melanoma Cells which Trigger (s) Hypoxic Response? Do Factors Associated with Hypoxia Affect Tumor Vessel Maturity? Is Hemorrhage the Result of Increased Immature Blood Vessel Density or, even worse, a Propensity of Melanoma Cells Themselves to Develop Immature Vessels through the process of Trans-differentiation? If T regulatory (FoxP3+) Cell Number is Not a Negative Prognostic Factor, What is the Role of Immune Checkpoint Proteins in Established Melanoma Brain Metastases? What is the Role of Brain Tumor Microenvironment in Mediating Immune Suppression or Blood Vessel Supply?
Hypothesis: Tumor Hypoxia Upregulates Angiogenic Factors and Abnormal (Immature) Blood Vessel Formation Carmeliet Nat Rev Drug Discov 2011
Methods-Approach Perform Tumor Imaging Analysis of Craniotomy Specimens from Patients with Melanoma Brain Metastases Stain (IHC) for C-inhibitory Immune Checkpoint Proteins (PD- L1, Galectin 9), and BV Density (CD31) Blood vessels (BV) were further defined by their maturity on the basis of expression of a pericyte marker, α-smooth muscle actin (SMA) Each stain was individually quantified in 4 tumor tissue compartments by a neuropathologist who set electronic gates from scanned slides in Aperio: melanoma cells, reactive glia, normal brain, and lymphocytic clusters. Areas with necrosis and hemorrhage were excluded The OS information was used to define the optimal cut-point in H-score between high - vs. low -expression of each biomarker
Quantification of Immunohistochemical Stains
Immunohistochemical Expression of PD-L1 in Melanoma Brain Metastases
Aperio Analysis Workflow (1) Tissue Segmentation Define Electronic Gates Corresponding to Different Tumor Tissue Compartments (PD-L1 stain example) Normal Reactive Glia Lymphocytic Clusters Melanoma Tissue
Aperio Analysis Workflow (2) Color Deconvolution & Calculation of H-score (VEGF example)
Detection/Quantification of Blood Vessel Density and Maturity Assumptions Only vessels with >50μm 2 diameter were included in analysis Maturity was defined by the presence of pericytes (SMA positive) Use of Definiens Architect with Tissue Studio Portal Armulik Dev Cell 2011
Definiens Architect Workflow (1) Definition of Mature-Immature Blood Vessels SMA positive area SMA alone SMA + CD31 SMA negative area CD31 alone
Definiens Architect Workflow (2); From A to Z Raw Image ROI Detection 1 3 4 2 2 3 1 Blue, vessels with lumen Red, vessels without lumen 5 1. SMA(+) region; 2. SMA( ) region; 3. Slide folds; 4. Slide artifact; 5. Glass 4 5 Yellow, small Orange, medium Red, large
Different Tissue Compartments Invariably Contribute to PD-L1 and Angiogenic Cytokine Expression 41 18 11 18 44 15 7 13 45 19 12 17 43 22 2 17
No Significant Differences Among Different Compartments in Blood Vessel Density and Quality 41 9 4 41 9 4
Prognostic Significance-Hypoxia/Angiogenesis Use of Survival Information to Define Optimal Cutpoints for high vs. low p=0.025 p=0.06 VEGF Melanoma N=44 Optimal H-score cutpoint: 10 High Low p=0.026 p=0.08 CD31 + SMA - Melanoma N=41 Optimal vessel density cutpoint: 75/μ 2 High Low
Prognostic Significance (2)-PD-L1 Use of Survival Information to Define Optimal Cutpoints for high vs. low p=0.031 Time from craniotomy to death (days)
Correlation Analysis Confirms Several Prior Known Associations bfgf CD31 HIF-1α PD-L1 VEGF IBV MBV Imm Infi Hemor bfgf 1 +0.33* -6.0** +6.9** CD31 1 +0.44* +2.4** HIF-1α 1 +0.63** +0.59** PD-L1 1 +0.36* -0.45* -18.5* VEGF 1 IBV 1 +0.33 +38.7* MBV 1 Imm Infl 2 Hemor 2 1 Pearsons Correlation Coefficient Pairs of Same-Patient Observations 2 Generalized estimating equations (GEE) models Noman J Exp Med 2014 Shi Pathology 2007 Ang Nat Gen 2002
Conclusions Tissue Imaging Analysis Supervised by Pathology Input Can Allow for More Objective Quantification of Protein Expression in Different Tissue Compartments Brain Tumor Microenvironment (reactive glial and/or immune clusters) Can Contribute to Immune Modulation and Response to Hypoxia Despite this Study s Cause-and-Effect Limitations in Interpretation of Results, Response of Melanoma Cells to Hypoxia (e.g. HIF1α, VEGF) may Paradoxically have Good Prognosis (VEGF is a trophic factor for CNS?) Treatment Implications for Anti-angiogenic Strategies and Immunomodulatory Strategies for Patients with Active Melanoma Brain Metastases
Supported by the University Cancer Research Fund, UNC-CH University of Pittsburgh Medical Center Ronald Hamilton MD-Neuropathology Michal Krauze MD, PhD-Hematology/Oncology Jonette Werley BSc-Neuropathology Stephanie Bortoluzzi-Neuropathology John Kirkwood MD-Medical Oncology University of North Carolina at Chapel Hill (UNC-CH) Yuri (Dimitri) Trembath MD, PhD-Neuropathology Evan Bradler BSc-UNC-CH School of Medicine Shanti Rao BSc-UNC-CH School of Medicine Angelica Saada BSc-SUNY Downstate School of Medicine Ana Snavely PhD-Biostatistics Nana Feinberg-Nikolaishvilli PhD-Translational Pathology Laboratory Bentley Midkiff, BSc-Translational Pathology Laboratory