Cytology meets Molecular:
Don’t throw away your microscope!
Andrew Fischer, M.D.
Director of Cytopathology
University of Massachusetts
British Association
for Cytopathology
November 4, 2017
Theme: How will molecular
testing and cytology evolve • What can molecular testing replace, using next
generation sequencing and bladder cancer
screening as an example.
• What molecular testing cannot do.
• The criteria of malignancy, when viewed from the
proper perspective, provide an irreplaceable insight
into the function of oncogenes.
• New microscopy techniques—fluorescent optical
sectioning and live cell imaging—will keep our
cytomorphologic skills in demand.
Cost in US (medicare):
~40$
Sensitivity for high
grade UC: Up to about
90%*
Sensitivity for ALL UC,
high and low grade,
<50%
High grade urothelial carcinoma, “microbiopsy” cytology sample
from catheterized ureter
Lee et al, Causes of False-Negative for High-Grade Urothelial Carcinoma in Urine Cytology.
Diagn. Cytopathol. 2016;44:994–999
Cost of Next generation sequencing: As low as 15,000 base pairs per penny
3 billion basepairs in whole
human genome
Only 1.5% is exomic
Need 30X coverage
(average 30 reads per DNA
molecule)
Costs approaching $1000
for biggest machines.
= ~15,000 bp for 1 cent
https://www.nature.com/news/technology-the-1-
000-genome-1.14901
Next generation sequencing for
urothelial carcinoma screening
• Compared to germline sequencing (two alleles, so 50%
of sequences are from one allele) NGS needs many
individual reads of the sequence to detect tumors that
are diluted with many normal gene sequences.
• Many genes are potentially mutated in cancers, so you
need to sequence many segments of DNA to have a
high sensitivity (~600 segments, ~100,000 bp). Would
cost (best case scenario) at least $150.
• A DNA methylation survey identified only ~150 DNA
segments, preferentially methylated in cancer, that may
be more suitable.
For NGS to work, need a high proportion of abnormal
cells. NGS could not be effective for this case.
In general, most molecular tests cannot work if the input material
is not assessed. Cytology is ideal for assessing input material,
because so few cells (~5000) are needed.
UroMark NGS platform for detecting high and low grade
bladder cancers
•Detected bladder cancer
with sensitivity of 98%,
specificity of 97%.
•To detect 1% tumor cells,
would need very roughly
2000 reads for each
amplicon (20 individual
tumor DNA sequences).
•Assume each amplified
sequence is 100 base
pairs.
•2000 X 150 X 100 /
15,000 = $20.00
• NGS-based testing is poised to replace at least some
urine cytology as a screening test for cancer.
• Likely to evolve through a phase of reflex testing for
cytologic atypia when percent abnormal cells is high.
• Most likely to first replace cytology for follow-up of
bladder cancer patients with known mutations.
Will Molecular take over urine cytology?
96.5 % of driver
genes for PTC have
been identified.
TCGA has not shed ANY light on why PTC
has diagnostic changes in chromatin and
nuclear shape.
Does Cytomorphology matter?
Molecular
vs.
Cytology
MAP kinase pathway, From Wikipedia, accessed 10-30-17
From Luger et al., Nature 389:251, 1997
Structure of the nucleosome, showing ~150 nucleotides of
DNA wrapped around the 4 pairs of histones. Note the
protruding histone tails.
Normal ductal cells vs pancreatic IPMN
A structure-function relation should also exist
at the cellular level
Hypothetical phylogeny of a cancer
Fischer, Young, DeLellis, J Cellular Biochem 93:28-36, 2004
“Hallmarks of cancer” are the hypotheses
for how oncogenes work.
In fact, 35 years after the discovery of oncogenes, we
still do not know how they actually work!
Hanahan and Weinberg, Cell 100:57, 2000 Hanahan and Weinberg, Cell 144:646, 2011
Phylogeny of Darwin’s Finches
Fischer, Young, DeLellis, J Cellular Biochem 93:28-36, 2004
Darwin knew that the morphologic
features that distinguish related
species…
…are ultimately caused by the heritable
elements (genes) that differ between the
species.
…relate in “the most essential yet often
hidden manner” to the mechanism of
increased fitness.
Darwin’s theory predicts there
should be a relation between the
morphologic changes diagnostic of
cancers, the genes that are active in
the cancers, and (most importantly)
the functional changes that allow
this cellular evolution to take place
ASC-sponsored Classification of
the Criteria of Malignancy
1. Tissue-level architectural changes reflecting clonal expansion.
2. Changes reflecting genetic instability
3. Sub-cellular changes, unrelated to genetic instability, conferring increased “cellular fitness”
A. Directly induced by oncogenes
B. Likely heritable, but no known genetic basis
Fischer et al., J Cellular Biochem 2010, 110:75-811
Group 1: Tissue architectural criteria Three common mechanisms for early clonal expansion of epithelial cells
“Crowding”
“Papillary formation” “True
stratification”
Fischer, Fundamentals of Cytological Diagnosis and Its Biological Basis. In: Pathobiology of
Human Disease. Published by Elsevier, 2014. p. 3311-3344
Loss of contact
inhibition,
Pseudostatification
Anchorage-independent
growth, resistance to
anoikis
Columnar
cell change
breast
Normal
ductal cells
Papillary thyroid carcinoma (Photomicrographs courtesy of Manon Auger, McGill)
Hepatocellular carcinoma, defined by true
stratification away from endothelial cells.
CD 34 staining to show endothelial cells
The development of true stratification
typically coincides with loss of (“basal-
apical”) polarity. Stratified cells can
then orienting freely toward any
surface (cribriforming).
There is increasing evidence for
involvement of oncogenes in breaking
down normal “basal-apical” polarity
Mammary carcinoma, with randomized
positioning of secretions and random nuclear
axis.
Objective markers of
basal-apical polarity
would likely be useful for
diagnosis
Group 2: Cytologic changes reflecting
genetic instability
Cell to cell variation in:
• Total DNA content (total integrated amount of
hematoxylin staining), reflecting chromosome
instability
• Cytoplasmic features (evidence of phenotypic
instability)
• Chromatin packaging patterns (?Epigenetic instability?)
DNA content variation reflecting chromosomal
instability in two HSIL pap tests
Polyploidization is a common
benign change
Benign polyploidization compared to
malignant nuclear pleomorphism
Group 3: Cell structural changes, not
related to genetic instability*
• Nuclear shape abnormalities
• Intermediate filament organization
• Other cytoplasmic diagnostic changes
• Abnormal nucleolar prominence
• Chromatin alterations
*Several of these criteria are known to be directly
induced by the cancer genes active in the tumor
E4 gene of
HPV
mediates
intermediate
filament
collapse
Group 3: Intermediate filament abnormalities
Group 3: Collapse of intermediate filaments
(keratin 20) in Merkel cell (small cell) carcinoma
Normal prostate epithelium
Prostatic intraepithelial neoplasia:
Diagnostic feature is nucleolar
prominence without reactive
cytoplasm
Fischer et al. (J Cellular Biochem, 2004)
Papillary thyroid carcinoma
Papillary thyroid carcinoma compared to normal thyroid
Normal thyroid epithelium
Follicular adenoma Papillary thyroid carcinoma
RET/PTC
TRK/PTC
B-RAF
H-RAS et al.
Thyroid model of carcinogenesis
Ret/PTC expressed in normal human thyroid cells
Fischer et al, Am J Pathol 153:1443, 1998
Normal human
thyroid epithelium
Human thyroid epithelium
expressing RET Tyrosine
kinase
Human thyroid epithelium
expressing H(V14)- RAS
Nuclear lamina
irregularity is
induced by
RET/PTC in
interphase.
Cells were micro-
injected with RET/PTC
6 hours previously and
stained for lamins
(green) and RET (red)
Fischer et al., Am J
Pathol 163:1091, 2003
Active RET/PTC
Inactive RET/PTC
(control)
Waggle dance of the
honeybee:
• A distinguishing trait from closest relative—the less social bumble bees*
• Different “dance dialects” have a genetic basis, shown by back-crossing to be defined by a single locus*
• Likely to be a key determinant of the fitness difference that allowed their evolutionary split.
• Essentially impossible to decipher this biology based on just the DNA sequence, and impossible to decipher with just snap shot images.
*R. N. Johnson, B. P. Oldroyd, A. B. Barron, and R. H. Crozier, J Heredity, 93:170-3, 2002
From Dan Schmolze et al., Arch Pathol Lab Med,
135:255, 2011.
Two photon microscopy
Right top: pseudocolored DAPI and fluoresceein to look like
H&E. Bottom right: H&E stained paraffin section.
• Living human papillary thyroid
carcinoma from thyroidectomy
specimen.
• Vibratome sectioned at 300
micron thickness.
• Stained for 5 minutes with 0.01%
acriflavine.
• 800 nm 2 photon excitation.
• Optically sectioned about 60
microns in depth.
• Viewed in just one time point.
• Diagnostic features of PTC
are seen in LIVING CELLS
Ending Comments • Molecular diagnosis will replace some cytologic tests. However, it
will mostly increase the role for cytology, through an increased use
of minimally sized biopsies that are ideal when molecular and
cytology are used together.
• Oncogenes are selected for at the cellular level. The DNA level of
biology CANNOT by itself inform us about how oncogenes actually
work.
• An increasingly refined understanding of the cytologic criteria of
malignancy brings us closer and closer to a description of the
exact functional changes in cancer cells.
• New microscopy techniques make it feasible to visualize new types
of diagnostic “cellular dynamics”, dynamics that will require our
expertise for use in diagnosis and prognostication for generations
to come.