Future of Molecular Pathology

Centro de Estudios Biosanitarios, Madrid, Spain, Virtual Class,

November 23, 2020

Ignacio I. Wistuba, M.D

Division Head ad interim, Division of Pathology and Laboratory Medicine.
Professor and Chair, Department of Translational Molecular Pathology
Co-Director, Khalifa Institute for Personalized Cancer Therapy (IPCT)
The University of Texas MD Anderson Cancer Center, Houston, TX.

USO EXCLUSIVO MOM

 Disclosures

• Advisory Board: Genentech/Roche, Bayer, Bristol-Myers

Squibb, Astra Zeneca/Medimmune, Pfizer, HTG Molecular,
Asuragen, Merck, GlaxoSmithKline, Guardant Health,
Oncocyte, and MSD.
• Speaker: Medscape, MSD, Genentech/Roche, Platform
Health, Pfizer, AstraZeneca, Merck
• Research support: Genentech, Oncoplex, HTG Molecular,
DepArray, Merck, Bristol-Myers Squibb, Medimmune,
Adaptive, Adaptimmune, EMD Serono, Pfizer, Takeda,
Amgen, Karus, Johnson & Johnson, Bayer, Iovance, 4D,
Novartis, and Akoya.

USO EXCLUSIVO MOM

 Paradigms in Cancer Molecular
Pathology - 2020

• Histology subtyping of cancer is still clinically important.

• Multiple clinically relevant molecular abnormalities (“driver
alterations”) have been detected and can be used to direct
targeted therapy and improve patients’ outcomes.
• Liquid biopsy represents an alternative option for molecular
testing, and potentially, early diagnosis.
• Immunotherapy-related biomarkers are part of diagnosis ii]n
cancer but predictive biomarkers are needed.
• Digital pathology, image analysis, quantitative imaging and
machine learning are emerging fields in cancer.

USO EXCLUSIVO MOM 3

 Bending the Curves?

Survival %

Biopsy/Blood
Time

USO EXCLUSIVO MOM

Lung Cancer Targeted Therapy Landscape 2018

Progression Free Survival

USO EXCLUSIVO MOM

R S Herbst et al. Nature 553, 446–454 (2018)
Paradigms in Cancer Molecular
Pathology – Beyond 2020

• Molecular diagnosis
• Immune diagnosis.
• Digital pathology.
• Integrative approaches

USO EXCLUSIVO MOM 6

 Diagnostic Algorithm for Lung Cancer
Diagnosis 2020

cfDNA Testing
(Liquid Biopsy)

Biopsy Cytology Blood

SCLC LCNEC Squamous Adenoca NSCLC-NOS

Morphology Morphology Morphology Morphology

Morphology
IHC NE (+) IHC p63/p40 (+) IHC TTF1 (+) IHC (-)

EGFR and BRAF mutation, ALK,

ROS1 and NTRK fusions, MET ex14
splicing

PD-L1 IHC
USO EXCLUSIVO MOM
 Molecular Testing for NSCLC - 2020
Adenocarcinoma
AKT
BRAF VEGFR
ALK HER2

Traditional RET
ROS1
EPHA/B
PDGFR
NTRK FGFR
Unknown
INSR

EGFR PI3K

Squamous MAPK
KRAS
Adenocarcinoma pG12C

Squamous Cell Ca
Large Cell

FGFR1
Amp

EGFRvIII
Adapted from W. Pao and N Girard, Lancet Oncol, 2011 Unknown PI3KCA
EGFR TK
DDR2
USO EXCLUSIVO MOM
 Genomic Abnormalities in
Lung Adenocarcinoma

Early Stage Metastatic Stage

USO EXCLUSIVO MOM

F. Skoulidis, Cancer Discovery, 2019
 Mechanisms of Resistance to EGFR TKIs

Acquired Resistance Acquired Resistance

Acquired First-Generation
Resistance First-Generation AcquiredThird-Generation
Resistance Third-Generation

USO EXCLUSIVO MOM

T. Nagano et al, Cells, 2018 Nov 15;7(11)
 Comparison of NGS with Conventional
Sequencing Technologies

Platform Sensitivity Sample Multiplexing Throughput Type of Quantitative

(for Requirement Capability Changes
clinical
use)

Sanger 20% High None Low SNV, No

indel
Pyro- 5% Intermediate None Low SNV Yes

Sequenom/ 5-10% Intermediate Intermediate Medium SNV Yes

ABI
SNaPshot

NGS 1-5% Low High High SNV, Yes

(amplicons indel,
and samples) fusion,
CNV
SNV: single nucleotide variations
CNV: copy number variations

USO EXCLUSIVO MOM

Indels: insertions/deletions
Courtesy of Raja Luthra, PhD, MD Anderson Cancer Center
 Next-Generation Sequencing (NGS)
Panel - Major Benefits
• Provide information in multiple targetable gene
abnormalities.
• Data on mutation, copy number variations, indels and
translocations
• Can be performed in routine small FFPE tissue samples and
liquid biopsy (cfDNA, CTCs, exosome DNA).
• Turn around time acceptable for clinical management and
costs being significantly reduced.
• Clinically, it offers to patients more options to get off-label
treatment and enter in genomic-based clinical trials.
• May provide information on tumor mutational burden (TMB),
and immune-suppressive genotypes (e.g., LKB1 mutations)

USO EXCLUSIVO MOM

MD Anderson New NGS Panel Content

613
Genes
Somatic Alterations Covered Others

613 Single Nucleotide Variants 126 MDACC actionable genes

33 Fusions 38 MSI genes/loci

613 CNVs and Indels 50 Top DDR pathway genes

TERT TERC non-coding MET exon

Promoter gene 14 skipping

Tumor Mutational Burden Estimation

USO EXCLUSIVO MOM

 Liquid Biopsy in Lung Cancer

• Currently, it is used in metastatic disease to deliver

targeted therapy:
• Can be easily repeated to control treatment efficiency and/or
the detection of genomic changes resulting from resistance to
therapy (e.g., EGFR T790M)
• It is an alternative to patients with solid tumors when biopsies
are inaccessible or after more than one attempt the yield was
unsatisfactory

• Other applications:
• Tumor mutational burden (TMB)
• Monitoring response to immunotherapies
• Minimal residual disease (MRD)
• Early detection
USO EXCLUSIVO MOM
 Sources of Liquid Biopsy

Enrichment

Isolation

Quantification

USOS.EXCLUSIVO MOM
A. Joosse and K. Pantel, Cancer Cell, Volume 28, Issue 5, 2015, 552–554
 Characteristics and Terminology for
Circulating Tumor DNA (ctDNA)

167 bp fragments of DNA, a nucleosome

Normal
cells/tissue

Circulating cell-free
DNA
cfDNA, ccfDNA

The linker DNA between nucleosomes is cleaved leaving 167

bp cell-free DNA fragments (145 bp plus a ~20 bp segment
ctDNA Tumor wrapping histone H1). Originally described by Wyllie in 1980.

Chandrananda et al. 2015 BMC Medical Genomics.

Wyllie 1980 Nature

USO EXCLUSIVO MOM

Slide courtesy of Dr. Rick Linman, MD
 cfDNA Genotyping Analysis
Blood cfDNA

Pre-analytical Next Generation of Sequencing (NGS)

Issues
Large/
Intermediate
Panels

Ion Proton Next-Seq

(Thermo Fisher) (Illumina)

Plasma
PCR-base Methods

• Amount of
blood/plasma Small Panels/
• Type of tubes Single Genes
• Time for
processing
Didigal Droplet qPCR
USO EXCLUSIVO
(dd)PCR (Biorad) MOM
(Cobas)
 Molecular Testing in Patients with Progression
or Recurrent Disease During Treatment with TKI

USO EXCLUSIVO MOM

J Thorac Oncol. 2018 Sep;13(9):1248-1268
 ctDNA Utility in Under-Genotyped
Non-squamous NSCLC
Tissue Biomarker
Positive
Tissue Biomarker
Positive 252, 19%
Tissue QNS/UG,
Tissue QNS or UG 383, 30% ctDNA Pos 383, 30%

Tissue QNS/PG,
Tissue Fully 879, 68% ctDNA Neg 627, 49%
Genotyped,
Biomarker Negative Tissue Fully
26, 2%
26, 2% Genotyped,
Biomarker Negative

ctDNA NGS Increased

Tissue Genotyping Biomarker Yield by
Status 65%
383 of 1288 (30%) ctDNA analysis
Biomarker Positive identified 252
additional actionable
879 (68%) Quantity biomarkers (19% of
Insufficient (QNS) 1288)
or Undergenotyped not previously
(UG) detected in tissue
QNS/UG cases

USO EXCLUSIVO MOM

Slide courtesy of David Gandara, MD, UC Davis, USA Zill et al, CCR 2018
 cfDNA Assay fo Lung Cancer Using
Centrifuged Supernatants from FNAs
Mutations
• From 150 lung cancer FNAs, all
Supernatant Tissue cfDNA
cases yielded enough DNA and 104
(90%) provided successful results by
NGS and ddPCR.
• Somatic mutations were detected in
82% of samples and relevant clinical
mutations in 50%.
• There was a high concordance
between mutation profiles of 67 cases
with tissue available: 100% with
concurrent FNA tissue/cells and 96%
with core needle biopsies (96%).
• In 45 cases with plasma cfDNA
samples concordance of driver
mutations was 84%

USO EXCLUSIVO MOM

Hannigan, et al, Annals of Oncology, 2019
 Paradigms in Cancer Molecular
Pathology – Liquid Biopsy

• Monitoring response to therapies

• Minimal residual disease (MRD)
• Early detection

USO EXCLUSIVO MOM 21

 Paradigms in Cancer Molecular
Pathology – Beyond 2020
Whole Exome Sequencing
and RNA Sequencing
(Bulk Tumor Tissue)
Single Cell Sequencing

A. Sudhagen, In J Mol Sci,, 2018

USO EXCLUSIVO MOM 22
cfDNA Analysis - Minimal Residual Disease

1
“Tumor-educated”
cfDNA Testing

2 3
Targeted Whole Patients Follow-up
Sequencing Exome Seq

USO EXCLUSIVO MOMModify from RI Chin, Mol Diag Ther, 201923

Paradigms in Cancer Molecular
Pathology – Beyond 2020

• Molecular diagnosis.
• Immune diagnosis.
• Digital pathology.
• Integration of data.

USO EXCLUSIVO MOM 24

Developing Markers for Immunotherapy
Topalian et al., 2012, NEJM
Herbst et al., 2014, Nature
PD-L1 IHC Garon et al., 2015, NEJM
Weber et al., 2015 Lancet
Taube et al., 2014 CCR
TILs Tumeh et al., 2014, Nature
Le et al., 2015, NEJM
Seiwert et al., 2015, ASCO
Th1/IFN-𝝲 Prat et al., 2017, Can Res
Ayers et a., 2017, JCI Regulatory
Phenotype
Serum IL-8 Sanmamed et al., 2017, Ann Oncol
approved as
markers Carleton et al., 2018 ASCO

T-cell prolif & MDSCs

Kitano et al., 2014, CIR
Huang et al., 2017, Nature
biomarkers
Sharma et al., 2018, AACR

Hugo et al., 2016, Cell

by the FDA
IPRES/Serpinb9 Pan et al., 2018, Science

Vetizou et al., 2015, Science

Microbiome Sivan et al., 2015, Science
Gopalakrishnan et al., 2018, Science

FDA
MSI Le et al., 2015, NEJM
Overman et al., 2017, JCO approved
Snyder et al., 2014, NEJM

Mutational burden
Van Allen et al., 2015, Science
Rizvi et al., 2015, Science
tests
Hugo et al., 2016, Cell

Genomic Mut.
Zaretzky et al., 2016, NEJM reporting
Gao et al., 2016, Cell

markers B2M/PBRM1/LKB1. Gettinger et al., 2017, Can Discovery

Pan et al., 2018, Science these metrics
Miao et al., 2018, Science

DNA FISH Ansell et al., 2014, NEJM

TCRβ clonality Tumeh et al., 2014, Nature

Robert et al., 2014, CCR

USO EXCLUSIVO MOM

Slide Courtesy of Dr. Kurt Schalper, Yale Cancer Center
 Multiscale and Dynamic Atlas of
Immune Changes
Comprehensive Longitudinal Patient-centric

Relapse
On-
treatment Post-
treatment
Baseline

USO EXCLUSIVO MOM

Modified from slide courtesy Sacha Gnjatic, Mount Sinai, New York, CIMAC
 Immune-profiling Workflows
Stools and other samples S16 sequencing

serum
100 µl
What is the
patient’s overall Luminex serum cytokines
inflammatory
state/immune
competence?
PBMC
106
Patient

FACS/Mag
Purified Cells CyTOF Hi-D Flow ATACseq
of Interest
Tumor

Single-cell TCRseq
Multiplexed Imaging/MIBI RNA Seq

What is the MIBI

immune state tumor
of the tumor? biopsy
of the tumor? biopsy
immune state tumor
What is the MIBI

Bulk or Single-cell TCRseq

WES, neotantigens

USO EXCLUSIVO MOM

Modified from slide courtesy Holden Maecker, Stanford University CIMAC
Cancer Immunity Cycle and Immune Checkpoints

USO EXCLUSIVO MOM

Adapted from Chen DS, et al. Immunity. 2013;39(1):1-
 Assessment of Immune Cells in Tumor
Tissues
Tonsil

Epithelium

Germinal Center

Adapted from Sid P. Kerkar, and Nicholas P. Restifo

Cancer Res 2012;72:3125-3130
USO EXCLUSIVO
PD-L1 MOM
CD3 CD8 CD4 CD68 DAPI AE1/AE3
 Immunopathology Laboratory
Translational Molecular Pathology – MD Anderson Cancer Center
Automated Digital Image Data
IHC/IF Pathology Analysis Analysis

Aperio AT2™
Genie™
Aperio/Genie™

Leica™

Inform™
Vectra/Inform™
USO EXCLUSIVO
E. Parra MOM
Vectra/Polaris™ and Luisa Solis, MD Anderson Cancer Center, 2018
 PD-L1 expression in NSCLC (H-score)
PD-L1 (3+) PD-L1 (2+) PD-L1 (1+) PD-L1 (0+)

E1L3N®
(Cell
Signaling)
200µm 200µm 200µm 200µm

TAIC expression in NSCLC (Cell Density; 9 markers)

PD-1 CD3 (pan T) CD4 (T helper) CD8 (T Cytotoxic) CD45Ro (T memory)

200µm 200µm 200µm 200µm 200µm

CD57 (NK) Granzyme B (NK & cytotoxic) FOXP3 (T reg) CD68 (macrophages)

200µm 200µm 200µm

200µm

USO EXCLUSIVO MOM

Edwin Parra et al, Clinical Cancer Res, 2016
 Types of Tumor Microenvironment in NSCLC
146 Adenocarcinomas (ADC) /108 Squamous Cell Carcinoma (SCC)
PD-L1 >5% / Intratumoral TAICs+ (CD3/CD68) moderate and severe density

PD-L1 + Adaptive Immune

Resistance
Immunological
Ignorance
PD-L1 -
TAICs + TAICs -
ADC 20% ADC 30%
SCC 26% SCC 28%

PD-L1 - PD-L1 +
TAICs + TAICs -
ADC 47% Tolerance Intrinsic
ADC 3%
SCC 40% (other suppressors?) Induction
SCC 6%
Michele W.L. Teng et al. Cancer Res 2015;75:2139-2145

USO EXCLUSIVO MOM

Edwin Parra et al, Clinical Cancer Res, 2016
Immune Checkpoint Expression in NSCLC
PD-L1 B7-H3 B7-H4

IDO-1 ICOS VISTA

LAG3 OX40 TIM-3

USO EXCLUSIVO
Edwin Parra andMOM
Pamela Villalobos et al, J Thoracic Oncol, 2018
Immune Checkpoints (N=9) IHC Expression in
NSCLC (N=184 Cases)

Malignant
PDL-1

Cells
B7-H3
B7-H4
IDO-1
PDL-1
B7-H3

Tumor Associated
Immune Cells
B7-H4
IDO-1
Vista
LAG3
TIM-3
ICOS
OX-40

Adenocarcinoma Squamous Cell Carcinoma

High (> median) Low (< median)

USO EXCLUSIVO
Edwin Parra andMOM
Pamela Villalobos et al, J Thoracic Oncol, 2018
 Multiplex Immunofluorescence (mIF) for
Tissue Immune-profiling

Opal™ - TSA System for Polaris™ InForm™

Multiplex Polaris IF (Perkin Elmer) (Perkin Elmer)

Edwin Parra-Cuentas,
MD, PhD

AE1/AE3 PD-L1 B7-H3 B7-H4 IDO-1 OX40 VISTA

(Human Tonsil) (Human Tonsil) (Lung ADC) (Prostate) (Human Tonsil) (Human Tonsil) (Human Tonsil)
IHC
IF

USO EXCLUSIVO E.MOM

Parra, et al, MD Anderson Cancer Center, 2018
 Multiplex IF VectraTM Panels
Panel 1 Panel 2 Panel 3 Panel 4 Panel 5

TILs & PD-L1/PD1 Immune Checkpoints Myeloid Cells

PD-L1 CD3 PD-L1 LAG-3 CD11b

PD-1 CD8 IDO-1 TIM-3 CD14

CD3 Granzyme B B7-H3 ICOS CD33

CD8 CD45Ro B7-H4 OX-40 CD66b

CD68 FOXP3 Vista CD3 CD68

-- -- CD3 CD20 CD163

AE1/AE3 AE1/AE3 AE1/AE3 AE1/AE3 AE1/AE3

DAPI DAPI DAPI DAPI DAPI

USO EXCLUSIVO MOM

E. Parra et al, MD Anderson Cancer Center
 mIF Vectra/Polaris™ Segmentation and
Co-localization Analysis

mIF Segmentation Panel 1 Co-localization:

PD-L1
Multiplex Segmentation Phenotyping Malignant Cells Macrophages
Cells PD-L1 +
IF AE1/AE3 + CD68 +
Panel 1

PD-L+ MCs PD-L1- MCs PD-L1+

AE1/AE3 CD3 TAMs
Tumor Stroma T cells CD4/CD8- Helper T Cells
Helper T Cells Cytotoxic T Cells
TAMs PD-L1+ TAMs PD-L1-
Malignant cells &
DAPI Malignant Cells Macrophages
Cytotoxic T Cells Macrophages
AE1/AE3 + / PD-L1 + CD68 + / PD-L1 +
PD-L1+
Panel 2

AE1/AE3 PD-1 Granzyme B MCs PD-L1- NK GraB/CD57

Tumor Stroma
Helper T Cells CD57 DAPI PD-L1 FOXP3
CD45RO/PD1
E. Parra et al, Science Reports, 2017
CD45RO Other cells
CD45RO

USO EXCLUSIVO MOM

E. Parra et al, MD Anderson Cancer Center
Immune Cells Infiltration in Chemotherapy-treated Lung Cancer

Multiplex Immunofluorescence (mIF) Increase of Immune Cells

Panel 1 Panel 2
Subpopulations
Non-small Cell Lung
Non-treated

Carcinoma (NSCLC)
Markers (N=112)
P-Value*
(cells/mm2)
Chemo Neo-adjuvant-
Naïve treated
(N=61) (N=51)
CD3+ 903.21 1501.99 0.021
CD57+ 206.87 527.35 <0.001
CD45RO+ 668.75 1180.26 0.019
CD45RO/PD1+ 153.73 443.04 <0.001
PD1+ 336.02 795.21 <0.001

Increase of Malignant Cells

Treated

PD-L1 IHC Expression

PDL-1 Expression (%)

P=0.008

PD-L1 CD3 CD8 CD4 CD57 GZB PD-1 FOXP3

CD68 AE1/AE3 DAPI CD45RO AE1/AE3 DAPI

USO EXCLUSIVO MOM

Chemo Naïve
E. Parra et al, JITC 2018
Treated
A Dormant TIL Phenotype Defines NSCLCs Sensitive
to Immune Checkpoint Blockers (anti-PD-1/PDL-1)
Multiplex Quantitative Immunofluorescence (QIF) Panel CD3/GZB/Ki67
(n=49 patients)

USO EXCLUSIVO MOM

Gettinger SN, et al. Nat Commun. 2018;9(1):3196.
Tissue Multiplex Immune Profiling Methodologies
Multiplex Immunofluorescence - 9 markers/panel
Vectra/Polaris™ InForm™
Opal™ - TSA System for
Multiplex Polaris IF (Perkin Elmer) (Perkin Elmer)

E. Parra, et al, Sci Rep. 2017 Oct 17;7(1):13380.

Imaging Mass Multiplexed Ion Nanostring GeoMxTM CODEXTM Digital

Cytometry Beam Imaging Digital Spatial Spatial Profiler
(CyTOF, Helios, Fluidigm) (MIBISCOPE, IonPath) Profiler

Cosma A, Cytometry A. 2017 Keren L, et al. Cell. 2018 Toki MI et al, Clin Cancer Res. 2019
Jan;91(1):12-13. 174(6):1373-87.e19.

USO EXCLUSIVO MOM

Tissue Multiplex Immune Profiling Methodologies

Type of Solid biopsy

sample (imaging system)

Light-based technology Mass-based

Approach technology

GeoMx CODEX MIBIScope Imaging Mass

(nanostring, (Akoya, CO- (IONPath, Multiplexed Cytometry*
Digital Spatial Detection by Ion Beam Imaging (Fluidigm, Imaging
Profiler) indEXing) system)
Technology Mass Cytometry
Hyperion system)

Application Discovery (screening) or exploratory analysis

Advantages Disadvantages
• High-plex capabilities (>10 markers) • Low acquisition speed
• High dynamic range • Low imaging throughput
• High dimensional data • Complex data analysis and visualization

*The image is acquired in the Flow Cytometry and Cellular Imaging Core Facility at MDACC

USO EXCLUSIVOAlejandro
MOM Francisco, MD PhD, MD Anderson Cancer Center
 Imaging Mass Cytometry (IMC) by Hyperion (Fluidigm)
Development of a Immune Panel (N=35 markers)
NKATPase Single Markers
GADPH
HLA-DR
B2MG
NKG2D/CD57
CD20 CD68
Granzyme B
CD3 CD137 CD163
CD94
CD8 TIM-3 Arginase 1
CD45RO VISTA
CD4 CD11b
OX40
ICOS CD14
PD-1 PD-L1 CD66B
FOXP3 CD73 IDO-1
LAG3
Ki67 B7-H4 CD33
B7-H3 Combined Markers
Histone H3 aSMA
Ir DNA-Intercalator CD31
Vimentin
Ruhtenium

panCK

LYMPHOCYTES

MYELOID CELLS

CARCINOMA CELLS Tonsil

STRUCTURAL AND STROMAL

Pedro Rocha, Alexandro Francisco et al , MD Anderson Cancer Center, AACR 2020

USO EXCLUSIVO MOM

Ir/CD3/CD8/Ki67 Ir/CD3/CD8/Ki67 CK
Imaging Mass Cytometry (IMC) by Hyperion (Fluidigm)
A Development of a Immune Panel (N=35 markers)

Tumor Tissue Segmentation

Tumor Tissue Profiling by IMC

Tumor Stroma

Spatial Analysis

Iridium Cytokeratin HLA-DR

Ki67 CD8

USO EXCLUSIVO MOM

Pedro Rocha, Alexandro Francisco et al , MD Anderson Cancer Center, AACR 2020
 GeoMxTM Digital Spatial Profiler Workflow

Antibody Imaging and region of UV exposure and Digital

incubation interest selection oligo collection Quantification

Leica
Bond Rx DSP
nCounter

USO EXCLUSIVO MOM

Luisa Solis, MD, MD Anderson Cancer Center
 GeoMxTM Digital Spatial Profiler Workflow

* * * * * * * * * *

CD20

CD20

Conventional IHC Nanostring DSP

• Good correlation of differentially expressed molecules between platforms

• Responders versus non-responders
Amaria, R.N., Reddy, S.M., Tawbi, H.A. et al. Neoadjuvant immune checkpoint blockade in high-risk resectable melanoma.
Nat Med 24, 1649–1654 (2018).

USO EXCLUSIVO MOMSlide courtesy of Dr. Michael Tetzlaff

CODEX (CO-Detection by Indexing) Profiling for Single
Cell Spatial Analysis

• Each antibody conjugated to a unique

oligonucleotide barcode
• FFPE/Fresh Frozen tissue stained with entire
antibody panel
Autostainer Scanner
• During each imaging cycle, three Reporters (dye-
labeled oligonucleotides) bind to their
complementary Barcodes and emit a
fluorescence signal upon excitation.

USO EXCLUSIVO MOM

Images shared by Akoya Biosciences
 Two Facets of Spatial Processes
Spatial Statistics and Mathematical Models

Several statistics and models can be applied to understand spatial point processes

Spatial Functions Description

Intensity Mean intensity (# of points/area) aka density
Nearest neighbor Distance to the nearest neighbor of each cell
K-function Expected number of cells around any random cell
L-function Transformation of the K-function to accommodate linear
scale
G-function Nearest neighbor distance distribution function
F-function Free space function
J-function Ratio of (1-G) and (1-F)
Hazard rate Ratio of derivative of F and (1-F)
PCF Pair correlation function - derivative of K-function
Tstat Third order statistics
Relative risk Segregation patterns for marks

Spatial statistics

• Depends on the window of observation

Spatial stochastic processes
• No mathematical modeling for this
presentation

Lavanya Kannan, Tarjani Agarwal, Matija Snuderl, David Zagzag, Erik Sulman, Jason Huse, and

USO EXCLUSIVO MOM

Kasthuri Kannan. Gibbs Process Determines Survival and Reveals Contact-Inhibition Genes in
Glioblastoma Multiforme. bioRxiv 608414; doi: https://doi.org/10.1101/608414
 Spatial Immune Infiltration
Distribution Metrics
• Cluster Correlation Functions CD3+/CD3+CD8+

pair correlation(r)
PD-L1-

PD-L1+

Dispersed Random Clustered

r
• G-Function Estimations For Macrophages

Macrophages/CD68+

r
PD-L1+ Macrophages tend to cluster in
Random PD-L1+ tumors while they are
PD-L1- dispersed in PD-L1- tumors
USO EXCLUSIVO MOM
K. Kannan, MD Anderson Cancer Center, 2020
 Segregation And Density Estimation

How fast does cell proliferation happen for a P2 CIM 13_[35015,12942]

particular phenotype?c

(Relative Risk Estimation ) Do cells of a

particular phenotype proliferate faster than any
other type?

Relative measure of proliferation is given by

relative probability of different types of
phenotypes
Cytotoxic T-cells Tregs
Epidemiological settings have case-control
studies. Relative risk is given by:
lD (u)
r (u) =
lC (u)
Similarly,

𝛽𝐶𝐷3+𝐶𝐷8+ (𝑢)
𝜌 𝑢 =
𝛽(𝑢)

USO EXCLUSIVO MOM

K. Kannan, MD Anderson Cancer Center, 2020
Paradigms in Cancer Molecular
Pathology – Beyond 2020

• Molecular diagnosis.
• Immune diagnosis.
• Digital pathology.
• Integration of data.

USO EXCLUSIVO MOM 50

 Digital Pathology Applications

• Clinical pathology diagnosis and biomarker

assessments (HER2, Ki-67, PD-L1, etc.).
• Quantitative Image Analysis to identify new
diagnostic, prognostic and predictive biomarkers:
machine learning and artificial inteligence
 Determination of molecular subtypes
 Analysis of the microenvironment tumoral
• Integration with clinical, genomic and immune
response data

USO EXCLUSIVO MOM 51

Digital Pathology Applications
Tissue Segmentation

Clinical Outcome

Cell Type Identification

USO EXCLUSIVO MOM S. Wang, EBioMedicine, 2019

52
Computational Pathological Image Analysis
and Lung Cancer Prognsosis
Image and Data Processing (n=48 features)

Non-small Cell Lung Cancer Prognosis (TCGA) Adenocarcinomas

N=523
Squamous Cell
Carcinomas
N= 511

USO EXCLUSIVO MOM

X. Luo et al. JournalThorac Oncol, 2017
 Digital Pathology Applications – Lung
Cancer Tumor Microenvironment (TME)
Prognostic value of the 48 TME feature–based
Nuclei Segmentation prognostic model in 371 NSCLC.

Correlation Between Image

Features and mRNA Expression

USO EXCLUSIVO MOM

Shidan Wang et al. Cancer Res 2020;80:2056-2066
Diagnostic Algorithm for Lung Cancer
Diagnosis – Beyond 2020
Pathology Clinical Informatics
Digital Pathology

Whole Exome Sequencing

RNA-Sequencing
Single Cell Sequencing

Liquid Biopsy “Tumor-educated”

Sequencing
USO EXCLUSIVO MOM
Paradigms in Cancer Molecular
Pathology – Beyond 2020

Tumor Microenvironment

Molecular

Pathology

USO EXCLUSIVO
Pathology MOM
Clinical Informatics 56
 Acknowledgments
Translational Molecular Pathology Collaborators
Tissue Banking ITH Projects, Lung MIRA CPRIT Grant
Junya Fujimoto, MD, PhD Jay Zhang, MD, PhD
Gabriela Mendoza Andrew Futreal, PhD
MDACC Institutional Tissue Bank, Dipen Maru, MD, Elena Bogatenkova Jack Lee, PhD
Cesar Moran, MD
TMA Sets and clinical database Neda Kalhor, MD
Junya Fujimoto, MD, PhD Stephen Swisher, MD
Carmen Behrens, MD Paul Scheet, PhD
Beatriz Sanchez-Espiridion, PhD
Immuno-profiling
IHC/IF and Digital Pathology Lab Don Gibbons, MD, PhD
Edwin Parra, MD, PhD Jack Lee, PhD
Luisa Solis, MD Chantale Bernatchez, PhD
Hitoshi Dejima, MD Cara Haymaker, PhD
Gabriela Raso, MD Boris Sepesi, MD
Alejandro Francisco, MD, PhD
Mei Jiang Immune Response & Genomics LKB, EMT & WES
Pedro Rocha, MD Ferdinando Skoulidis, MD, PhD
Barbara Mino John Heymach, MD, PhD
Don Gubbons, MD, PhD
Genomic Studies Roy Herbst, MD, PhD
Ximing Tang, MD, PhD Jack Lee, PhD
Dzifa Duose Stephen Swisher, MD
Wei Lu
Chi-Wan Choi
Tina McDowell
Humam Kadara, PhD

MD Anderson APOLLO program

Andrew Futreal, PhD
Grace Matthew, PhD
Celia Garcia-Prieto, PhD
USO EXCLUSIVO MOM
Thank You

USO EXCLUSIVO MOM