Deciphering the biology that drives response to immunotherapy Phenoptics TM Quantitative Pathology Platform Trent Norris, Field Application Scientist September 15, 2016 HUMAN HEALTH ENVIRONMENTAL HEALTH 2014 PerkinElmer
Triple Negative Breast Cancer Survivor My Aunt Full Remission Since 2014
Triple-negative breast cancer: Range of histology. Clifford A. Hudis, and Luca Gianni The Oncologist 2011;16:1-11 2011 by AlphaMed Press
Triple-negative breast cancer: Recurrence and survival. Clifford A. Hudis, and Luca Gianni The Oncologist 2011;16:1-11 2011 by AlphaMed Press
Cancer Immunotherapy is providing lasting benefits Personalized immunotherapy is the future Anti-PD1s Leading the way The immune system is the agent that improves outcome and cures people with metastatic solid cancer.
Immuno-oncology is not yet personalized PD-L1 IHC helps pick out responders, But PD-L1 but negatives response for positives averages 40% also respond, thus making test of little use
https://www.statnews.com/pharmalot/2016/08/05/bristol-myers-opdivo-fails-lungcancer/
The biology that drives response is a lot more complicated Response depends on specific cell-to-cell interactions
Cell-to-cell signaling - markers of interest
Melanoma example, conventional PD-L1 IHC How can we characterize cell-to-cell interactions with IHC?
Making immunohistochemistry precise and quantitative Visual Protein Assessment Quantitative, multi-analyte, per-cell needs new imaging and staining methods The tissue is the issue Still the gold standard primary diagnosis and most directly connected to disease Information not presently being accessed with conventional approaches Samples are getting smaller and less available
Phenoptics Quantitative Pathology Solutions Reagents Instruments Software Services Opal multiplexed immunofluorescence staining kits and protocols Vectra Automated Quantitative Pathology Imaging Systems (6-slide and 200-slide) Mantra Quantitative Pathology Workstation inform Tissue Finder advanced image analysis & phenotyping software TIBCO Spotfire multivariate analytics and data analysis Full workflow services: Method development Slide staining and scanning Data analysis Assay transfer
Reagents: Sequential application of colors TSA enables use of multiple antibodies raised in the same species in the same tissue section (or even the same cell compartment) MW Tissue section Antigen A Antigen B Antigen C Antibody A (Rabbit) Antibody B (Rabbit) Antibody C (Rabbit) HRP-conjugated Anti-Rabbit IgG or Superpicture Fluorochrome tyramide
Advantages of Opal TM 1. Select antibodies based on performance, not species 2. TSA enables 1. Amplifies the signal so camera exposures of 10s of ms vs 100s or 1000s 1. Key for automated scanning, and for clinical workflow 2. Ease of balanced signals goal of factors of 2 or 3 for best unmixing 3. Reduce photobleaching can scan slides multiple times with <10% reduction in signal 3. Can multiplex up to 6 markers, possibly increasing in the future. 4. Less use of expensive antibodies because of high signal:noise 5. Assay optimization is faster & more predictable than conventional multicolor indirect immunofluorescence (weeks versus months)
Melanoma example, conventional PD-L1 IHC
Phenoptics multiplexed IHC - Opal red = PD-L1 yellow = CD8 green = Foxp3 magenta = CD20 aqua = sox10 pink = CD163 blue = DAPI
Analyzing samples stained w/ many colors Multiplexed stained samples are difficult to interpret Colors blend together and are difficult to resolve Expression patterns are hidden by colocalization Visual assessment is qualitative at best Autofluorescence is also present in FFPE samples Solution: Multispectral imaging separates multiple markers trainable pattern recognition automates segmentation of images into tumor and stroma, collects per cell data for phenotyping, retains spatial distribution information Monochrome Color (RGB) Multispectral
Viewing 10x whole slide views in Phenochart
Simulated IHC views for Opal TIL kit applied to breast cancer CD4 Opal520 CD20 Opal540 CD8 Opal570 Foxp3 Opal620 CD68 Opal650 CK Opal690 DAPI Spectrally separated H&E view Foxp3 view Cytokeratin view CD68 view CD20 view CD8 view CD4 view
inform learn-by-example interface
inform learn-by-example interface
inform learn-by-example interface
inform learn-by-example interface
Phenoptics multiplexed IHC - Opal red = PD-L1 yellow = CD8 green = Foxp3 magenta = CD20 aqua = sox10 pink = CD163 blue = DAPI
Phenoptics multiplexed IHC - Opal With tumor detection and cell phenotyping Tissue segmentation red = tumor PD-L1 green = stroma yellow blue=background = CD8 green = Foxp3 red = PD-L1+ tumor cell magenta aqua = PD-L1- = tumor CD20 cell green = regulatory T cell aqua sox10 yellow = cytotoxic T cell pink magenta = CD163 = B cell pink = macrophage blue DAPI Phenotyping blue = other
Example of Quantifying PDL1
Example #1 PD-L1 simulated IHC
Example #1 PD-L1 scoring Tissue segmentation red = tumor green = stroma blue=background Scoring blue = 0+ yellow = 1+ orange = 2+ red = 3+
Phenoptics multiplexed IHC - Opal red = PD-L1 yellow = CD8 green = Foxp3 magenta = CD20 aqua = sox10 pink = CD163 blue = DAPI
Breast Cancer Example Dr. Beth Mittendorf, MD Anderson
Case #1 pseudo composite cyan = CK purple = killer T cell green = helper T cell red = B cell orange = PD-L1 yellow = Foxp3
Case #1 inform tumor-stroma and cell phenotype maps Phenotype Counts Tumor 4,380 Phenotypes tumor killer T helper T T-reg B cell other Killer T 146 Helper T 640 Regulator T 156 B cell 208 Other 2,500 Total 9,260
Case #1 - Spatial point pattern analysis in R (example calc.) 33% of tumor cells have a killer T cell within 25 microns distance
Case #2 with tumor / stroma map and cell phenotypes Phenotype Counts Tumor 1,810 Phenotypes tumor killer T helper T T-reg B cell other Killer T 213 Helper T 0 Regulator T 14 B cell 1,270 Other 2,160 Total 6,467
Case #3 with tumor / stroma map and cell phenotypes Phenotype Counts Tumor 4,630 Phenotypes tumor killer T helper T T-reg B cell other Killer T 1,070 Helper T 962 Regulator T 236 B cell 80 Other 1,230 Total 8,206
Case #4 with tumor / stroma map and cell phenotypes Phenotype Counts Tumor 3,620 Phenotypes tumor killer T helper T T-reg B cell other Killer T 203 Helper T 285 Regulator T 127 B cell 218 Other 1,850 Total 6,307
Case #5 with tumor / stroma map and cell phenotypes Phenotype Counts Tumor 2.940 Phenotypes tumor killer T helper T T-reg B cell other Killer T 2,570 Helper T 1,670 Regulator T 192 B cell 61 Other 1,540 Total 8,968
From strong interaction to weak 61% 41% 33% 3%
AE
Tissue Map Filter Settings merged_cell_seg_data - Slide ID: (FIHC4_Multi_11)
Scatter plots and Histograms
Future directions laying groundwork for clinical use Digital pathology workflows Autostaining Fluorescence whole slide Simulated IHC region selection multiplex analysis 20x whole slide image hematoxylin and DAB Future developments under QSR Slide analysis times getting significantly shorter Developing protocols for Bond Rx Process time for 7-color assay approx. 14 hours Today Field Level Analysis Phenotype Spatial / R-scripts Informatics Solutions Case Level Analysis Case Study Level Analysis
PD-1 Blockade with Pembrolizumab in Advanced Merkel Cell Carcinoma Merkel cell carcinoma 56% response with Merck Pembro viral status, mutation burden, PDL1 IHC, CD8 infiltration, not predictive Phenoptics revealed immunobiology consistent with anti-pd1 MOA. Potential basis of a predictive test
Novel quantitative multiplexed PD1/PDL1 IHC test provides superior prediction to treatment response in melanoma patients Conv. PDL1 IHC has a PPV of ~30% Multiplexed IF PPV 79%, NPV 83%
I think we are making progress! What do you think? Questions? red = PD-L1 yellow = CD8 green = Foxp3 magenta = CD20 aqua = sox10 pink = CD163 blue = DAPI
We re all in it to win it together! Thank you Trenton Norris trenton.norris@perkinelmer.com