Pathology Terminology

1
Pathology Terminology

• Issues and Solutions

2
Cells are constantly responding to changes in
their LOCAL environment by transcription.Gene
microarrays provide a snapshot of the
transcriptional cell environment of one cell type
(cell culture) or the average of a population of
cells (tissue, organ). Out of context (no
knowledge of environment or cell types) it means
nothing. Context is provided by pathophysiology,
timecourse and measured parameters.
3
At study completionFour fundamental questions of
a microarray experiment

• What makes up my sample?
• Gross anatomy, histology, histopathology,image
  analysis, ultrastructural pathology,
  immunohistochemistry, ISH
• What is the local environment?
• Histopathology, cell biology, phys and
  biological chemistry
• Where am I in the chronology of the lesion?
• Pathology
• How did it get that way? Mechanism or
  pathogenesis
• Physiology,pathophysiology, molecular
  pathology,biological chemistry, physics and
  physical chemistry, gene expression,
  proteomics,genomics,metabonomics, nutrigenomics

4
Both the pathologist and the molecular biologist
must start to think in terms of macro and micro
environments and the measurement and interaction
of the two ---- and since tissue morphology is
dynamic, no single ontology will work for all
situations.
Define your sample by its environment
5
Organism environment -age, sex, energy sources
(food composition), time of sampling, conditions
of sampling, treatment
Location Environment organ, substructure(lobes),t
issue
Cellular Environment-histology/pathology,
chemistry, physical
Subcellular environment-organelles,ultrastructural
pathology
Molecular environment
Gene expression
6
Factors which will alter gene expression patterns
Age, sex, strain(race) Time of food
consumption Diet defined-protein, carbs,
fat(PUFA,HUFA, Sat)-beef v fish based Concomitant
Medications Clinical findings-hypertension Pre-te
rminal medications/acid/base/fluids/blood
gases/radiation Significant inter-organ
lesion/interaction Time to sample
collection Known SNPs or sample
translocations/mutations(REDPATH) Location of
sample--blood in sample or perfused? Composition
of sample-(eg. liver portal triads,
fibrosis) Active processes---(fibrous connective
tissue present but fibroplasia in progress or
not?) Concurrent infections-HIV,
HPV,HCVetc Menstrual/estrus stage/neurohormonal
factors Clinical chemistry and hematology-glucose,
lipids, eosinophilia Diagnostic criteria and
terminology-because that sets cohort
7

• The technical side of microarrays is fairly
  well established.
• - most critical factors experimental design,
  sampling and cohort selection.
• Since cellular environments will determine gene
  expression, should histopathology be a
  requirement for defining cohorts? YES
• What information will be needed for a complete
  histologic and pathologic picture?
• - Who will decide?
• - How will samples be curated?
• - How will databases be standardized?

8
I can select my own samples. Why do I need a
pathologist?
Even if you get 80-90 agreement on a term, the
local environment heterogeneity for the same
lesion may be enormous. Steatosis may be local,
lobular, diffuse, macrovesicular, microvesicular,
both graded, not graded, included in cirrhosis,
not included in cirrhosis,.associated with
fasting, associated with high fat feeding,
associated with ligaments.
9
If you ask 3 pathologist to diagnose the same
sample, You might get 3 different answers!
Actually, you get one or two answers, it just
sounds like 3 answers
This is partly because of highly flexible
nomenclature Often necessitated by highly
flexible biology
10
Why is nomenclature an issue in pathology?

• Descriptive interpretive basis-Virchow
• Classifications and gradings based on outcomes
• Interpretation may be based on one 4m section
  out of a 20 cm organ
• Relationships of the entire organism taken into
  consideration
• i.e. uremic mineralization in the stomach
• One persons experience may influence his
  diagnosis
• Terms may be redundant but not necessarily
  synonymous

11
Pathologists have spent years placing
samples into predictive cohorts
Cellular morphology and tissue structural
characteristics form the foundation for biologic
behavior and outcomes----dont throw away 200
years of experience that will allow the proper
selection of cohorts
12
Examples of tumor histology impacting gene
expression
I have a 1 gram piece of tissue from a human
brain tumor. I cut off 100mg sections and place
them in Trizol with no histology. The diagnosis
from an adjacent piece of tissue is in the
database I simply pull this sample out by the
annotation. Everytime I repeat the hyb with the
same sample I get a slightly but often
significantly different answer. Why?
13
Case kindly provided by Dr. Sydney
Finkelstein-Redpath Integrated Pathology
14
AREA 1
AREA 2
15
The sample of the tumor on the left had not only
a more aggressive morphology but had accumulated
4 LOH mutations and differentially expresses
genes not noted in the portion on the
right-- --this is lost under the term
glioma -staging would help
16
This is a case of one tumor with different
morphologies within the tumor, different
mutations impacting the gene expression and
probable different impacts on local gene
expression
From a 4u section, how do we know what the sample
looks like 1 mm deeper in the tissue? Highly
dependent upon gross sampling. Step section
image analysis with heterogeneity score Bob
Maronpot-NIEHS, NTP
17
Histologic Variation of Normal----food
effects- What constitutes Normal?
18
Variable size brown and white adipocytes in
inguinal fat of db/ Mice
WAT
WAT
WAT
BAT
WAT
19
Uniform large white adipocytes in db/db inguinal
fat
WAT
WAT
WAT
WAT
WAT
WAT
WAT
WAT
WAT
20


All of the above are downregulated in adipose
tissue by a high fat diet
21
Proteins are only part of the story
Glucose sensitivity is dependent upon lipid
saturation index of cell membranes
So how many genes respond to glucose? .a bunch
Dietary lipids will be reflected in cell
membranes in 3 days to 3 weeks-we only measure
proteins but lipid-induced functional
consequences for proteins is huge.
22
Simply the process of collecting the sample is a
problem--- gene changes can occur within 10-15
minutes for biopsy material For postmortem
samples, what are the most labile
transcripts-depends on tissue?
Authors Nishizawa T. Tamaki H. Kasuga N. Takekura
H. Title Degeneration and regeneration of
neuromuscular junction architecture in rat
skeletal muscle fibers damaged by bupivacaine
hydrochloride. Source Journal of Muscle Research
Cell Motility. 24(8)527-37, 2003. Abbreviated
Source J Muscle Res Cell Motil. 24(8)527-37,
2003. Publication Notes The publication year is
for the print issue of this journal. NLM
Journal Code hsn, 8006298 Journal
Subset IM Country of Publication Netherlands Abs
tract Degeneration of muscle fibers and NMJs was
observed 4 h after BPVC injection.
23
PCA-db/db mouse vs normal wildtype
MuscleHeart
Liver
Kidney
Muscle
Why the huge Variability in normal muscle?
Chart from Lei Zhu, Ph.D., Principal
Statistician Statistical Data Sciences, GSK
24
Part of difference may be fiber type differences
within the muscle
Both samples are from the same gastrocnemius
muscle- The only answer for this may be LCM
25
How precise does sample location designation have
to be? Liver, median lobe, 5cm from hilus, 2cm
from gallbladder--- depends on context
In Real Estate and Gene Expression its the
same.. Location, location, location
26
Acetaminophen Study Experimental Design
3 Time Points 6, 24, 48 Hrs following Single
Oral Dosing
50 mg/kg/day
6
24
48
0
150 mg/kg/day
1500 mg/kg/day
Time (Hrs)
Slides Kindly provided by Drs Malarkey, Boorman
and Maronpot NIEHS
NIEHS ToxPath Team
27
Tissue Collection Liver
Caudate
RP
Left
RA
Median
Slides Kindly provided by Drs Malarkey, Boorman
and Maronpot NIEHS
28
Acetaminophen hepatoxicity
There will be more globin genes here simply
because of the hemorrhage(RBCs)
Slides Kindly provided by Drs Malarkey, Boorman
and Maronpot NIEHS
29
Vascular pattern of necrosis
Slides Kindly provided by Drs Malarkey, Boorman
and Maronpot NIEHS
30
Number of Genes Differentially Expressed in all
Animals plt.005
Some toxicants or pathogens absorbed in proximal
small intestine impact median lobe Some toxicants
or pathogens absorbed from the colon impact the
left lobe
Irwin, et al. 2004. Tox Path (in press)
Slides Kindly provided by Drs Malarkey, Boorman
and Maronpot NIEHS
31
Same animals, same livers, different lobes
Left Lobe
Shunting of blood?
Median lobe
Irwin, et al. 2004. Tox Path (in press)
Slides Kindly provided by Drs Malarkey, Boorman
and Maronpot NIEHS
32
Dose and Time Considerations for Drugs
33
Predicting histopathology by microarray
Methapyrilene hepatotoxicity
This is the dose/time equivilent of the liver
lobe effect
Hamadeh,et al. 2002 Tox Path
34
Adaptative responses vs system overwhelmed and
cell death
The low dose allows for adapatation
Hamadeh,et al. 2002 Tox Path
35
Acetaminophen
Substrate/environ changes
Downstream gene targets
Extensive necrosis and loss of hepatocytes-fewer t
ranscribing cells
Fatty change ballooning degeneration
Hemorrhage, early necrosis
Heinloth, et al. 2004. Tox Sci
36
What is the best pathologic visualisation of
microarray data?
Begin with standardizaiton of cohorts-which means
standardization of nomenclature/diagnostic
criteria
37
Nomenclature(pathology) is not likely to be
standardized to the extent needed for
bioinformatics professionals to mine the data in
a meaningful way except in local
environmentsand even if it is standardized, its
use will always be suspect without seeing the
images(samples, slides).
38
Attempts to capture anatomy by ontology will be
by necessity overlapping and overly complicated
What is needed is a graphic interface (terms
running in the background) linked to a
committee-vetted and certified set of
images(samples) and histories. A percent fit
would be calculated for the morphology of each
image. Another percent fit would be calculated
for the other parameters and a total fit score
would be present for each comparison. Processes
comprising the image could also be scored eg.
apoptosis, cell proliferation
A visible man ( rat ,mouse, rhesus) approach
would allow detailed sample site identification
and the retrieval of thumbnail JPGs or TIFF files
from certified samples---animal models
included. Certified samples would be required in
a MIAME-like fashion for journal submissions and
storage into a national database.
39
Organism environment -age, sex, energy sources,
time of sampling, conditions of sampling,
treatment
Location Environment organ, substructure(lobes),t
issue
These require a fit algorithm within the
database
Cellular Environment-histology/pathology,chemistry
Subcellular environment-organelles,ultrastructural
pathology
Molecular environment
Gene expression
40
What is the best way to visualize the
morphologic sample cohort?
Clinical presentation
Annotate patient ID and Select sample site
a
Location Fiber type
Sample Histopathology- Cell types present
Local environment-pH, serum lipid composition,
creatine, lactate, glucose, hypoxia period at
collection?,temperature
41
Structure and ultrastructure compartmentalize
proteins for functional reasons----we should not
ignore this---an ontology based on structure and
substructure separates proteins into areas of
physical contact or proximity ----not an end but
a rational start.
42
The GO Consortium has already provided the basis
for this visual depiction of gene expresssion
The Gene Ontologies Molecular Function Ontology
the tasks performed by individual gene products
examples are carbohydrate binding and ATPase
activity Biological Process Ontology broad
biological goals, such as mitosis or purine
metabolism, that are accomplished by ordered
assemblies of molecular functions Cellular
Component Ontology subcellular structures,
locations, and macromolecular complexes examples
include nucleus, telomere, and origin recognition
complex
43
For the tissue and then for each cell type if LCM
and iSH is available and necessary
Organelles light up by number of genes
altered/out of total possible for that organelle
44
Sublocalization within the organelle if possible
45
Modified from BioCarta by Ellen Rosenbloom
46
1440302_at
1447856_x_at
Ca binding
1434210_s_at
1420836_at
1425013_at
1420967_at
1425948_a_at
1420835_at
1416345_at
Inner mitochondrial membrane
1416955_at
1438360_x_at
1436874_x_at
1434801_x_at
1430542_a_at
1438922_x_at
1438546_x_at
1438545_at
1438187_at
1424563_at
1434897_a_at
1423981_x_at
1423980_at
1423109_s_at
1424912_at
Oxalic,Malonic Succinic,Glutaric Adipic,maleic,fum
aric, Phthalicall dicarboxylic acids
ATP,ADP,AMP Energy balance
FA oxid
1424317_at
1437459_x_at
47
Mitochondrial matrix
FA oxidation 1415984_at acetyl-Coenzyme A
dehydrogenase, medium chain 1416408_at
acyl-Coenzyme A oxidase 1, palmitoyl 1416409_atac
yl-Coenzyme A oxidase 1, palmitoyl
1416090_atpyruvate dehydrogenase (lipoamide)
beta
TCA cycle 1416478_a_at malate dehydrogenase 2,
NAD (mitochondrial)1415891_atsuccinate-CoA
ligase, GDP-forming, alpha subunit
Amino acid metabolism 1416209_at glutamate
dehydrogenase
48
Functionality-link to Kegg/Biocarta/ Clinical
parameters
Peroxisome
1368232_atmevalonate kinase
Cholesterol biosynthesis
1367775_atalpha-methylacyl-CoA racemase
1369070_atPEX12
Bile acid metabolism
1368878_atisopentenyl-diphosphate delta isomerase
1367885_atPxmp2
Peroxisome biogenesis
1368264_atPex6
1368526_atPex3
1367995_atcatalase
Response to ROS/H202
1387215_atalanine-glyoxylate aminotransferase
Transaminase-alanine/glyoxylate Serine/pyruvate
1369663_atepoxide hydrolase 2, cytoplasmic
1368057_atATP-binding cassette, sub-family D
(ALD), member 3
ATPdep transport
Peroxisome Matrix
Aromatic metab and Pos reg blood pressure
1377887_atacyl-CoA oxidase 3, pristanoyl
1368150_atSlc27a2
Fatty acid transporter
1368426_atcarnitine O-octanoyltransferase
Fatty acid oxid
1368283_atenoyl-CoA, hydratase/3-hydroxyacyl CoA
dehydrogenase
1368890_atacyl-CoAdihydroxyacetonephosphate
acyltransferase
1386885_atenoyl coenzyme A hydratase 1
1387271_atphytanoyl-CoA hydroxylase
1368427_atA kinase (PRKA) anchor protein 11
Membrane
type II cyclic AMP-dependent kinase (PKA) response
49
(No Transcript)
50
Differential gene expressionapproach to
understanding disease
Generic cell death/inflammation responses
Disease
Normal
Initiating cell/tissue/organ
Secondary organs
Peroxisomes,lipid, Glycogen, Mitochondria
Causative
Related pathways
Structural Pathology Non-structural
Unrelated pathways
Plasma membrane Endoplasmic reticulum Ribosomes
Human samples near the normal end will require a
large heterogenous database that will by
necessity need this type of gold standard samples
Measurable parameter (serum vs cellular) Non-meas
urable?
Metabolic Osmoregulatory Hormonal Neural
Cell Signalling
Adaptive compensated
There may be no morphologic changes in this phase
but samples are essential
Adaptive decompensated with damage (adaptive for
abnormal/not normal)
Time
51
Summary

• Gene expression in context over time coupled with
  biostatistical tools and integrated
• analysis should be able to characterize and
  stage disease states over time
• and provide an adjunct to strict histopathology
  and insights into disease
• mechanisms.
• PCA and PLS-DA are capable of finding patterns in
  differential gene expression in tissue
• if the cohorts are properly defined but does not
  work if they are not.
• Defining proper cohorts will require consistent
  annotation of specimens, measuring environments
• and consistent histopathologic assessment of
  submitted samples.
• Since gene expression reflects local cellular
  environments under sytemic
• influences, the degree of control over tissue
  collection, location, timing,
• and other conditions cannot be overstated and
  may require a degree of
• harmonizaiton for optimal results.
• It is probable that a tried and true structural
  approach to gene expression, as
• in histopathology and ultrastructural analysis,
  will yield valuable insights and a graphic
• interface may be simpler than attempting to
  rationalize different ontologies.

52
Thank you! Acknowledgements
53
Acknowledgements

• Dr Sidney Finkelstein
• Dr Dave Malarkey
• Dr Bob Maronpot
• Dr Gary Boorman
• Dr Lei Zhu
• Dr Jim Butler
• Dr Steve M Clark
• Dr Ellen Rosenbloom
• Dr W. David Benton
• Dr Mike Waters (NIEHS)
• Dr Jennifer Fostel
• Dr Susanna Sansone
• Dr Kevin Morgan

54
Extra Slides
55
Phenotypic anchoring
Hamadeh,et al. 2002 Tox Path
56
Hepatocyte Necrosis, Apoptosis, and
degeneration Centrilobular to massive /-
neutrophils /- mononuclear cells regeneration
57
Gene Expression Changes suggestive of Stress
Implicated in Oxidative Stress or DNA Damage
Responses
58
Mitochondrial Damage after Exposure to 150 mg/kg
APAP
Heinloth, et al. 2004. Tox Sci