multiplexed imaging, pathology and personalized medicine: keeping sight of the forest
DESCRIPTION
Tremendous efforts to appreciate the complexity of cancer and the specific threats and vulnerabilities of individual malignant tumors are underway, involving large-scale and high-resolution DNA, RNA and protein characterization. Results of these investigations may shape the kinds of tissue-based analytical tools that anatomic pathology will be called upon to develop and deploy. Increasingly, these often-multiplexed panels of molecular assays prove to be complex, expensive, proprietary, and hard to independently validate. But what if, in addition, these tests turn out to be focused on the trees and in fact are missing the forest? The interaction between cancers (highly mutable in their geno- and phenotypes) and their tissue environment (more stereotyped) has proven to be complex and informative. Recent data from multiple investigators suggest that host factors rather than specific tumor characteristics may determine clinical outcomes. How should these perspectives be reflected in research and clinical practice, particularly with respect to pathology-based tools? Perhaps a new focus on “stromics” is in order. In addition, much of contemporary tumor profiling is undertaken in the context of current conventional or targeted therapies. What will happen if and when more general cancer treatments (equally applicable to tumors with differing tissue backgrounds and molecular phenotypes) become reality? Such developments may lead to simpler and lower-cost assays with improved impact on prognostic accuracy, treatment selection and outcome.TRANSCRIPT
Multiplexed Imaging, Pathology and Personalized Medicine: Keeping Sight of the Forest
Richard Levenson, MD
Dept. of Pathology and
Laboratory Medicine
UC Davis Medical Center
Themes
1.The stroma (not the epithelium) may be key (…“STROMICS”)
2.“Systems biology” is most relevant at structural, intercellular and tissue levels
3.Bonus: glimpse of novel structural (target: collagen) and multiplexed (target: heterogeneity) imaging technologies
Are we looking in the right places?
How might a focus on micro-environment and host factors affect diagnosis and treatment strategies?
Normal breast
Invasive squamous cell carcinoma
Gabriella dental x-ray and CT: loss of bone
Nov, 2011 Feb, 2012
14 months later…
Implications:
New therapies will need new companion diagnostics—or perhaps none…
Molecular phenotypes in spatial context can be informative
Coming up:
Multiplexed imaging
Selection of targets
12
How well could we understand the battle of Waterloo…if we blenderized the battlefield?
Spectral imaging variants
PARISS
NUANCE/VECTRALCTF
HSi-440cAOTF
RebellionIMS
Vectra: Evaluating EMT in FFPE-Tissue Sections
C. Hoyt, PerkinElmer, Inc.; C. Gagen & R. Wetzel, Cell Signaling Technology
DAPI E-cadherin vimentin
Raw color image
Immunofluorescence with quantum dots
DAPIPR (585-nm QDot)ER (655-nm QDot)Convert to brightfieldDAPI and autofluorescence
16
From Mouse to Microscope: Breast cancer and Cy5-anti-Her2 antibody
Courtesy Roche, Penzberg, Germany
17
Four nuclear signals can be distinguished using chromogens in brightfield
Hematoxylin
Ki67—red cell cycle
pHH3—brown cell cycle
CC3—gray antiangiogenesis met. suppressor
Great tools and reagents…
So what compartments and components should we be looking at?
(hint: stroma)
Nature Medicine, 2008
Top 200 gene expression levels most different between microdissected normal and tumor-associated stroma.
Automatic clustering gave these recurrence-free survival curves
26-gene stromal signature is independent of all major clinical and pathological variables:
• ER • HER2• lymph node status • age• grade• tumor size
NOT PREDICTIVE!
C-Path: ‘Omic’s approach to the analysis of tumor morphology
AH Beck, …, D Koller. Systematic analysis of breast cancer morphology uncovers stromal features associated with survival. Science Translational Medicine 2011. 3, 108ra113.
Seventy-gene prognosis signature Size Invasiveness gene signature Mastectomy Wound response signature ERBB2 molecular subtype Grade Basal molecular subtype ER Lymph node Genomic grade index Luminal A molecular subtype Luminal B molecular subtype Chemotherapy
NOTSIGNIFICANT:
Classic tumor characteristics are not predictive
Andy Beck
• C-Path score • Hypoxia • Age
What is significant:
Just 3 stromal features predicted outcome better than 8 epithelial features
Moreover:
Andy Beck
Untreated Treated
2009
Tumor heterogeneity: Stromal environment affects stemness and HER2-positivity
(Stem cell marker)
In this case, HER2- induction NOT due to gene amplification
Implications for HER2 IHC and ISH assays?
Wicha 2013
What about structural properties of the stroma?
Can we detect them in tissue slices?
Boyd et al. Breast Cancer Research 2011 13:223
OR 1 1.7 2.1 2.4 4.7
Mammographic density affects the risk of breast cancer
0 <10 <25
<50 <75 >75
28
Stromal rearrangement around cancer can reflect biological properties
Linear collagen strands oriented tangentially (0 degrees)
Linear collagen strands oriented at 90 degrees
Non-invasive tumor
Locally invasive tumor
(Second harmonic generation imaging)
Provenzano et al., BMC 2008
TACS-2 TACS-3
Locker and Segal commentary on Conklin et al, AJP, 2011
Risk independent of:GradeSizeAgeER/PR/HER-2node status
Conklin et al, AJP, 2011
DISEASE-FREE SURVIVAL
SHG: complex, expensive
Inexpensive collagen detection
Tissue microarray sample: low magnification
Orientation of fibers viewed with color map
Bile duct in cirrhotic liver
Sneak preview: Subcellular-resolution multiplexed MS
Conventional imaging mass spec: spatial resolution ~35 microns
(Images by Richard Caprioli, keynote speaker at a previous Tucson Symposium)
Sneak preview: Subcellular-resolution multiplexed MS
Hematoxylin and Ki-67 imaged via mass spec
Spatial resolution < 1 micron
Large field of view, high-resolution mass-spec image
So what’s coming?
•Stromics •“Companion diagnostics” regime not
going away …(so ER/PR/Her2 etc. will be around for a while)
•… but disruptive therapies will alter the diagnostic landscape
•Spatially resolved, highly multiplexed subcellular-scale molecular and structural profiling is feasible
•New tools--and new paradigms—should allow us to see the leaves, the trees and the forest.
Prognosis: encouraging
The Venezuelan Poodle Moth is a possible new species of moth discovered in 2009 by Dr. Arthur Anker of Bishkek, Kyrgyzstan.
Snapshot collagen birefringence image of an H&E-stained breast cancer sample
Notes
• Instead of using canonical cell surface markers to identify leukemia stem cells (LSCs), Lagadinou et al. apply a functional approach… of ROS measurements: …both novel and potentially of great utility.
• Cell Stem Cell. 2013 Jan 15.
• BCL-2 Inhibition Targets Oxidative Phosphorylation and Selectively Eradicates Quiescent Human Leukemia Stem Cells.
• The Bcl-2 gene has been implicated in a number of cancers, including melanoma, breast, prostate, chronic lymphocytic leukemia, and lung carcinomas.
• “We demonstrate here that it is feasible to eradicate resistant LSC populations by targeting their unique metabolic dependencies.”
Simple antigen-positivity measurements in bulk tumors may be simplistic
Yellow = HER2(+) stem cells (not bulk population)
Wicha, 2013
FT-SHG: tissue-scale featureschange optics, then:
P-SHG: molecular organization(6 measurements required)
N.B.: Half the malignant stroma is “normal,” and localized features important, so imaging is necessary
-1
0
1
2
3
4
5
6
7
0 10 20 30 40 50 60 70 80 90 100
Tumor Transplantation (days)
Tu
mo
r B
urd
en (
cm3 )
Left Inguinal (Control rat)Right Axillary (Control rat)Left Axillary (Control rat)Right Axillary (Cured rat)Left Axillary (Cured rat)
Death
Effect of immunophotonic therapy on primary and metastatic tumors in rats
Primary
Mets
CD8 cell attacking a tumor cell
Tumor Immunotherapy: Stage 4 breast cancer
Biology of tumor progression meets biology of aging