Numerical Methods in Cancer Models: Doron Levy

Motivation
Properties
numerical methods
Stability in the delays space
Numerical Methods in Cancer Models
Doron Levy
Department of Mathematics
and
Center for Scientific Computation and Mathematical Modeling (CSCAMM)
University of Maryland, College Park
Doron Levy Montréal, May 2013

Motivation
Properties
numerical methods
Plan
1 What is cancer?
2 Delayed di↵erential equations
3 Agent-based models
4 PDEs

Motivation
Properties
numerical methods
Delayed Di↵erential Equations in Cancer Models

Analysis & Numerical Methods
Doron Levy
Department of Mathematics
and
Center for Scientific Computation and Mathematical Modeling (CSCAMM)
University of Maryland, College Park

Motivation
Properties
numerical methods
Outline
1 Motivation
Cancer Immunology
Stem Cell Transplantation
2 Properties
Zero Crossings
Time Scales
3 Numerical Methods
4 Stability in the delays space

Motivation
Properties The immune response to leukemia
numerical methods Stem cell transplantation
Outline
1 Motivation
Cancer Immunology
2 Properties
Zero Crossings
Time Scales
3 Numerical Methods

Motivation
What is leukemia?
Normal cells: stem cells turn into

mature cells
Leukemia: A malignant
transformation of a stem cell or a
progenitor cell
Myeloid or Lymphocytic
Acute or Chronic

Motivation
CML
3 phases
Chronic: uncontrolled proliferation
Accelerated
Acute: Uncontrolled proliferations. Cells
do not mature
Philadelphia chromosome
Translocation (9;22)
Oncogenic BCR-ABL gene fusion
The ABL gene expresses a tyrosine
kinase. Growth mechanisms
Easy to diagnose
Drug targeting this genetic defect (a
tyrosine kinase inhibitor)

Motivation
Treating leukemia
Chemotherapy
Bone Marrow or Stem Cell transplant
Chemo + radiotherapy +
transplantation
Imatinib (Gleevec)
Molecular targeted therapy -
suppresses the corrupted control
system
$32K-$98K/year

Motivation
Problems with existing therapies
Remission vs. Cure: Can CML be cured?

Yes! but only with a bone marrow (or stem cell) transplant
Requires a (matching) donor
A risky procedure (+ unpredictable side e↵ects)
Imatinib? Does not cure the disease: stopping it causes a relapse
New Medical Data: There is an anti-leukemia immune response (Lee
lab)
The strength and dynamics of the specific anti-leukemia immune
response can be measured
Number of cells
Activity (count signaling molecules)

Motivation
The observed immune response
A di↵erent immune response for each patient. However:

At the beginning of the treatment: no immune response
Peak: around 6-12 months (after starting the drug treatment)
Later: waning immune response
Question:
What is the relation between the dynamics of the cancer, the drug, and
the immune response?

Motivation
A mathematical model
x2n
ry divide nτ
Leukemic ay (a’y ) Progenitor b y (b’y ) Differentiated c y (c’y ) Terminal p0e-cnCkC

sT T cells (T)
stem cell (y0) cell (y1) cell (y2) cell (y3)
d0 + qCp(C,T) d1 + qCp(C,T) d2 + qCp(C,T) d3 + qCp(C,T) dT
Ingredients:
Leukemia cells: stem cells, , fully functional cells
Mutations, Drug (Imatinib), Anti leukemia immune response
Kim, Lee, Levy: PLoS Computational Biology, ’08
Michor et al. (Nature 05). Cronkite and Vincent (69), Rubinow (69), Rubinow & Lebowitz (75), Fokas,
Keller, and Clarkson (91), Mackey et al (99,...), Neiman (00), Moore & Li (04), Michor et al (05), Komarova & Woodarz (05).

Motivation
Michor’s model + immune response
Cells without mutations:
ẏ0 = [ry (1 u) d0 ]y0 qc p(C , T )y0 ,

ẏ1 = a y y0 d 1 y1 qc p(C , T )y1 ,
ẏ2 = b y y1 d 2 y2 qc p(C , T )y2 ,
ẏ3 = c y y2 d 3 y3 qc p(C , T )y3 .
Anti-cancer T cells:
Ṫ = st dt T p(C , T )C + 2n qT p(Cn⌧ , Tn⌧ )Cn⌧ ,

X
p(C , T ) = p0 e cn C kT , C = (yi + zi ), Cn⌧ = C (t n⌧ ).

Motivation
Accounting for the immune response
0.06
No immune
0.05
response
Cell Concentration (k/µL)
0.04
Leukemia
0.03
0.02
0.01
T cells
0
0 10 20 30 40 50
Time (months)

Motivation
Stopping imatinib (simulation)
40
Leukemia
35
30
1000 x T cell
25 concentration
20
15
10
Imatinib removed
at month 12
5
0
0 5 10 15 20
Time (months)

Motivation
Cancer vaccines: a mathematical design

0.5 -5
SC 0
PC
log10[concentration]
0.45 Leukemia cells -5
-10
0.4 -10
0.35 -15
-15
0.3
Anti-leukemia T cells -20
0 5 10 15
-20
0 5 10 15
0.25
DC TC
0.2 Inactivated leukemia 0 5
log10[concentration]
0.15 cells (vaccines) -5 0
0.1 -10 -5
0.05 -15 -10
0 -20 -15
0 2 4 6 8 10 12 14 16 0 5 10 15 0 5 10 15
Time (months) Time (months) Time (months)
Inactivated leukemia cells

V̇ = dV V qc p(C , T )V + sv (t)
Anti-cancer T cells
Ṫ = st dt T p(C , T )(C + V ) + 2n p(Cn⌧ Tn⌧ )(qT Cn⌧ + Vn⌧ )

Motivation
Model populations
Host Cells
Cancer
Anti-donor T cells
General blood cells
Donor cells
Anti-cancer T cells
(cancer-specific)
Anti-host T cells
General blood cells

Motivation
Everything takes time

Motivation
Anti-Cancer T Cells
σ σ
TD /TC Interaction
TC /C Interaction
Ignore pT2C/C pT2D /TC Survive

kCTC kTDTC
TC
React pT1C/C x2n
pT1D /TC Perish
dTC
Proliferate nτ
Reload υ
qT1C/C Death
ρ qT2C/C Die or
Become anergic
qT3C/C

Motivation
Anti-Host T Cells
σ σ
TH /TD Interaction
TH /C Interaction
pT2H /C pT2H/TD
Ignore pT2D/TH Survive Ignore
React pT1H /C kCTH kTDTH Perish

pT1D/TH pT1H/TD React
Proliferate Proliferate
qT1H /C nτ nτ qT1H /TD
x2n x2n
ρ Reload υ TH υ
ρ
qT2H /C
Reload
qT2H /TD
qT3H/C
σ x2n No flow
kHTH qT3H/TD = 0
TH /H Interaction
pT2H /H nτ dTH
Ignore υ
Die or qT2H/H Killed by TD
React pT1H /H qT1H /H
become
anergic
ρ Death

Motivation
Anti-Donor T Cells
σ σ
TD /TH Interaction
TD /TC Interaction
pT2D/TC pT2D/TH
Ignore pT2H/TD Survive Ignore
React pT1D/TC kTCTD kTHTD Perish React

pT1H/TD pT1D/TH
Proliferate Proliferate
qT1D/TC nτ nτ qT1D/TH
x2n x2n
ρ Reload υ TD υ
ρ
Reload
qT2D/TC qT2D/TH
σ x2n No flow
kDTD qT3D/TH = 0
TD/D Interaction
nτ
Ignore pT2D/D υ dTD
No flow Killed by TH
qT3D/TC = 0 qT1D /D qT2D /D
React pT1D/D
ρ Death

Motivation
General Donor and Host Blood Cells
Stem Cells Stem Cells

SD SH
D H
pT1D/DkDTD pT1H/H kHTH
TD /D Interaction
TH /H Interaction
dD Perish dH Perish
ρ ρ
Death Death

Motivation
Cancer Cells
logistic growth
C/TC Interaction
C/TH Interaction
rC
pC/T
1
H kT C
H pC/T
1
C kT C
C
Perish Perish
C
Death rate is included
in net logistic growth term
ρ ρ
Death

Motivation
Equations...
dTH
= dTH TH kCTH kTD TH kHTH
dt
T /C T /T T /T
+ p2 H kC (t )TH (t ) + p2 D H p2 H D kTD (t )TH (t )
TH /H
+p 2 kH(t )TH (t )
+ 2n p
TH /C TH /C
1 q 1 kC (t ⇢ n⌧ )TH (t ⇢ n⌧ )
n TH /H TH /H
+2 p 1 q 1 kH(t ⇢ n⌧ )TH (t ⇢ n⌧ )
n TD /TH TH /TD TH /TD
+2 p 2 p 1 q 1 kTD (t ⇢ n⌧ )TH (t ⇢ n⌧ )
TH /C TH /C
+p 1 q 2 kC (t ⇢ )TH (t ⇢ )
TH /H TH /H
+p 1 q 2 kH(t ⇢ )TH (t ⇢ )
TD /TH TH /TD TH /TD
+p 2 p 1 q 2 kTD (t ⇢ )TH (t ⇢ ).

Motivation
Time Delays
Time for reactive T cell-antigen interaction = 5min

Time for unreactive interactions = 1min
Time for cell division = 0.5-1.5 day
T cell recovery time after killing another cell = 1 day

Motivation
Relapse
−6
x 10
2 1.2
General host cells H
Cell Concentration in 10 3 cells /µL

1 Cancer cells C
1.5
0.8
Anti−host T cells TH
1 0.6 eventually overwhelms TH
0.4
0.5
0.2
never goes to 0
0 0
0 20 40 60 80 100 0 200 400 600 800
Time in Days Time in Days

Motivation
Remission
−6
x 10
4 1.2
General host cells H Cancer cells C

3.5
1
3 Anti−host cells TH
0.8
2.5
2 0.6
1.5
0.4
1
0.2 C = 0 at time 25.7276
0.5
0 0
0 20 40 60 0 20 40 60

Motivation
Oscillations
A: Stable oscillation B: Unstable Oscillation

0.4
Anti−host T cells
Cancer cells
0.35
General host cells 1

0.3
0.8
0.25
0.2 0.6
0.15
0.4
0.1
0.2
0.05
0 0
0 500 1000 1500 0 500 1000 1500

Motivation
Extinction instead of stability
Without state constraint With state constraint

25 25
Anti−host
20 T cells TH Anti−host T cells TH

15 20
10
Cancer cells C
5 15
−5 10
Cancer cells C
−10 The value of C crosses 0

The value of C crosses 0
−15 at time 8.0192. 5 at time 8.0192, and does
not recover.
−20
−25 0
8 8.05 8.1 8.15 0 2 4 6 8

Motivation
Initial anti-host cells vs. initial host cells
Results up to time 2000

Successful
16 Unsuccessful
(cells/µL)
Unresolved
14
Higher initial host blood cell
concentrations improve the
Initial anti−host T cell concentration TH,0
12
10 chances of a successful cure

8
Greater initial anti-host T cell
6 concentrations slightly favor the
4 chances of cure
2
1 1.5 2 2.5 3 3.5 4 4.5 5

Initial host cell concentration H0 (10 3 cells/µL)

Motivation
Number of cell divisions vs. cancer growth rate
Results up to time 1000

6
Successful
Unsuccessful
5.9 Unresolved
5.8 A higher average number of T

cell divisions favor complete
Average # of T cell divisions n
5.7
5.6
remission
5.5
5.4
Higher cancer growth rate make
5.3 complete remission slightly
5.2 more likely
5.1
5
0.03 0.035 0.04 0.045 0.05
Cancer growth rate rC (1/day)

Motivation
Properties Zero Crossing
numerical methods Time Scales
Outline
1 Motivation
Cancer Immunology
2 Properties
Zero Crossings
Time Scales
3 Numerical Methods

Motivation
Zero crossing: Example
A simple DDE:
dx
= rx(t 1), x(t) = 1, t < 0.
dt
Solve: t 2 [0, 1).

dx
= rx(t 1) = r
dt
Then
x= rt + c = 1 rt.
If r > 0 then x(t) = 0 for T 2 [0, 1)

Motivation
Zero crossing: example
A simple DDE:
dx
= rx(t 1), x(t) = 1, t < 0.
dt
Proceed: t 2 [1, 2)
dx
= rx(t 1) = r + r 2t r2
dt
Then
r2
x =1 rt + (t 1)2 .
2

Motivation

A simple DDE:
dx
= rx(t 1), x(t) = 1, t < 0.
dt
The general solution: t 2 [n, n + 1)

n+1
X (t k + 1)k
x(t) = ( r )k
k!
k=0
Question:
For what r does that exist a T 2 [n, n + 1) such that
n+1
X (T k + 1)k
( r )k = 0?
k!
k=0

Motivation

A simple DDE:
dx
= rx(t 1), x(t) = 1, t < 0.
dt
The general solution: t 2 [n, n + 1)

n+1
X (t k + 1)k
x(t) = ( r )k
k!
k=0
Question:
For what r does that exist a T 2 [n, n + 1) such that
n+1
X (T k + 1)k
( r )k = 0?
k!
k=0

Motivation
Time Scales
A toy problem
xn+1 = (1 yn )xn
yn+1 = xn2 + k + yn
The map iterated twice
xn+2 = (1 + xn2 k yn )(1 yn )xn

yn+2 = ((1 yn )2 + 1)xn2 + 2k + yn

Motivation
Time Scales
A toy problem
xn+1 = (1 yn )xn
yn+1 = xn2 + k + yn
The map iterated twice
xn+2 = (1 + xn2 k yn )(1 yn )xn

yn+2 = ((1 yn )2 + 1)xn2 + 2k + yn

Motivation
This corresponds to...

Motivation
Properties
numerical methods
Outline
1 Motivation
Cancer Immunology
2 Properties
Zero Crossings
Time Scales
3 Numerical Methods

Motivation
Properties
numerical methods
Approach #1
A mesh in time that is based on the delay.

Numerical methods for ODEs that use the mesh points only
(multistep methods)
Example:
⇢
y 0 (t) = f (t, y (t), y (t ⌧ (t))) , t0  t  tf ,
y (t) = (t), t  t0 .
A set of meshpoints:
= {t0 , t1 . . . , tN = tf },
such that tn ⌧ (tn ) 2 .

Forward Euler: yn+1 = yn + hn+1 f (tn , yn , yq ), q < n.
Same idea with Adams-like methods, Heun, etc.

Motivation
Properties
numerical methods
Approach #2: Feldstein

Free the mesh selection from the delay
Use extranodal points for the approximation of the delayed term
y (t ⌧ (t)).
Example: (t0  ↵(t)  t)
⇢ 0
y (t) = f (t, y (t), y (↵(t))) , t0  t  tf ,
y (t0 ) = y0 ,
Assume h = (tf t0 )/m. For any tn = t0 + nh, define

✓ ◆
↵(tn ) t0
q(n) = floor .
h
A numerical method:
⇢
yn+1 = yn + hf (tn , yn , zn ),
y0 = y (t0 ),
where zn = yq(n) , i.e., a piecewise-constant approximation of y (↵(t)).
Alternatively, a piecewise-linear approximation:
zn = (1 r (n))yq(n) + r (n)yq(n)+1 .
Motivation
Properties
numerical methods
Approach #3: Bellman’s method of steps
Assume a constant delay. In the first interval [t0 , t0 + ⌧ ] the DDE has the
form: ⇢ 0
y (t) = f (y (t), (t ⌧ )),
y (t0 ) = (t0 ).
In the second interval [t0 + ⌧, t0 + 2⌧ ], define y1 = y (t ⌧ ) and
y2 (t) = y (t). Then:
8 0
>
> y (t) = f (t ⌧, y1 (t), (t 2⌧ )),
< 10
y2 (t) = f (t, y2 (t), y1 (t)),
>
> y1 (t0 + ⌧ ) = (t0 ),
:
y2 (t0 + ⌧ ) = y (t0 + ⌧ )
And so on...
For every timestep, a larger system. However, this system can be solved
using standard methods for ODEs.

Motivation
Properties
numerical methods
Approach #4: Methods based on continuous extensions
Continuous Extension: very low cost method to get an accurate

approximation of the solution at every point in the interval
⌘(tn + ✓hn+1 ) = n,1 (✓)yn + ... + n,in +1 (✓)yn in

+hn+1 (yn , . . . , yn in ; ✓, g⌘ , 0n ), 0  ✓  1.
where
g⌘ (f , y ) = f (t, y , ⌘(t ⌧ (t, y ))).
This is how every Matlab routine provides solutions at the sampled
points

Motivation
Properties
numerical methods
Approach #4: Methods based on continuous extensions
Consider the DDE

⇢ 0
y (t) = f (t, y (t), y (t ⌧ (t, y (t)))) , t0  t  tf ,
y (t) = (t), t  t0 .
Using continuous extensions, solving the DDE amounts to solving

the ODE:
⇢ 0
wn+1 (t) = f (t, wn+1 (t), x(t ⌧ (t, wn+1 (t)))) , tn  t  tn+1
wn+1 (tn ) = yn ,
where 8
< (s), s  t0 ,
x(s) = ⌘(s), t 0  s  tn ,
:
wn+1 (s), tn  s  tn+1 ,
and ⌘ is the continuous extension interpolant.

Motivation
Properties
numerical methods
Additional Reading
Bellen and Zennaro, Numerical Methods for Delay Di↵erential

Equations, Oxford
Shampine and Thompson, Numerical Solution of Delay Di↵erential
Equations

Motivation
Properties
numerical methods
Outline
1 Motivation
Cancer Immunology
2 Properties
Zero Crossings
Time Scales
3 Numerical Methods

Motivation
Properties
#ROSSINGSET7;SQRT SQRT SQRTSQRT =^;=
numerical methods
DXDT DXDTDXT´ T DTDXT´T DTX T T


!
#

T

d 2x dx dx(t ⌧1 ) dx(t ⌧2 )

!+ + + + 8x = 0

´
dt"2 dt dt dt

´

´
T

DXDT DXDTDXT´ T DTDXT´T DTX T T T T

DXDT
DXDTDXT´
DXDT T DTDXT´T
DXDTDXT´ T DTDXT´T
DTX
DTX X

T T
T T
5NMARKEDAREASINTHEFIGUREABOVE

"

REPRESENTSTABLEREGIONS

!

4HENUMBERSDENOTETHENUMBEROF

! !

#

EIGENVALUESINTHERIGHTHALFPLANE

# #

.OTICETHAT#REPRESENTSANISOLATED

STABLEREGION

T
T

´

´
´

! !
´ ´
´
´

"LUELINESDENOTEBRANCHESCORRESPONDING ´
´

T
" " ´

TOINDICESOFTYPEUV

´ ´
´

'REENLINESDENOTEBRANCHESCORRESPONDING
T T

TOINDICESOFTYPEU V
T T T T
X

T T
T T
X

X
"

REPRESENTSTABLEREGIONS " "#



STABLEREGION
´ STABLEREGION
STABLEREGION
´ ´

´
´ ´
´
´
´
"LUELINESDENOTEBRANCHESCORRESPONDING ´ ´
TOINDICESOFTYPEUV "LUELINESDENOTEBRANCHESCORRESPONDING
"LUELINESDENOTEBRANCHESCORRESPONDING
TOINDICESOFTYPEUV
TOINDICESOFTYPEUV
´
´
´ ´
TOINDICESOFTYPEU V 'REENLINESDENOTEBRANCHESCORRESPONDING
TOINDICESOFTYPEU V
TOINDICESOFTYPEU V
T T
T T
T T

# Doron Levy Montréal, May 2013
Motivation
Properties
numerical methods
Stability crossing curves
Ref:
Gu, Niculescu, Chen, On stability crossing curves for general systems with
two delays, J. Mathematical Analysis & Applications, 311 (2005), pp.
231–253.
Stability crossing curves: The set of delays for which the

characteristic equation has at least one imaginary zero (or pair of
imaginary zeros).
Associated with change of stability

Motivation
Properties
numerical methods
The two delays case
A DDE with two constant delays
x(t) + c1 x(t ⌧1 ) + c2x(t ⌧2 ) + c3 x 0 (t) + c4 x 0 (t ⌧1 ) + c5x 0 (t ⌧2 ) = 0.
The characteristic equation:

⌧1 s ⌧2 s
h(s) = h0 (s) + h1 (s)e + h2 (s)e .
Let ak (s) = hk (s)/h0 (s). Then
⌧1 s ⌧2 s
a(s, ⌧1 , ⌧2 ) = 1 + a1 (s)e + a2 (s)e = 0.
Stability: a question of the number of the roots of the characteristic

equation with a real part on the right hand side of the plane.

Motivation
Properties
numerical methods
The two delays case

For an imaginary s = i! to satisfy a(s, ⌧1 , ⌧2 ) = 0, the vector
corresponding to the three terms must form a triangle:
⌧1 s ⌧2 s
a(s, ⌧1 , ⌧2 ) = 1 + a1 (s)e + a2 (s)e = 0.
For an imaginary s = i to satisfy
the vector corresponding to the three terms must form a triangle.
Hence, their magnitudes must satisfy the triangle inequalities:
Hence, their magnitudes must satisfy the triangle inequalities:
|a1 (i!)| + |a2 (i!)| 1,

1  |a1 (i!)| |a2 (i!)|  1.

Motivation
Properties
numerical methods
The two delays case
The triangle inequalities determine which i! may be zeros of a(s)

The set of all such ! are the crossing set ⌦.
Any given ! defines a collection of pairs (⌧1 , ⌧2 ).
\a1 (i!) + (2u 1)⇡ ± ✓1

⌧1 = 0, u = u0± , u0± + 1, . . .
!
\a2 (i!) + (2v 1)⇡ ⌥ ✓2
⌧2 = 0, u = v0± , v0± + 1, . . .
!
where from the law of cosine:
✓ ◆
1 + |a1,2 (i!)|2 |a2,1 (i!)|2
✓1,2 = cos 1 ,
2|a1,2 (i!)|
and u0± , v0± are the smallest possible integers such that the
corresponding ⌧1,2 are nonnegative.

Motivation
Properties
numerical methods
The two delays case

• The triangle inequalities determine which i may be zeros of a(s).
• We call the set of all such the crossing set .

• Any
The given set
crossing defines
• The⌦crossing
alwaysasetcollection
consists
always
ofofpairs
consistsof aa
( 1, number
finite
finite number
)
of2 intervals of intervals of
finite length of finite length .
• Any interval of ’s defines a collection of curves in R2.
• The
Any curvesofglue
interval together
!’s(symmetrical,
defines aincollection
different arrangements,
of curves in Rdepending
because imaginary roots come in conjugate pairs) 2 on
which triangle inequalities correspond to the endpoints of the
Theintervals.
general case is a union of the following sets:

Motivation
Properties
numerical methods
Example
DDE:
dx
Regions are marked with
= 2x(t ⌧1 ) + x(t ⌧2 ).
dtthe number
of zeros in right half plane.

Numerical Methods in Cancer Models: Doron Levy

Uploaded by

Copyright:

Available Formats

Numerical Methods in Cancer Models: Doron Levy

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Numerical Methods in Cancer Models: Doron Levy

Uploaded by

Copyright:

Available Formats

Motivation

Numerical Methods in Cancer Models

Doron Levy Montréal, May 2013

Doron Levy Montréal, May 2013

Delayed Di↵erential Equations in Cancer Models

Doron Levy Montréal, May 2013

4 Stability in the delays space

Doron Levy Montréal, May 2013

4 Stability in the delays space

Doron Levy Montréal, May 2013

Normal cells: stem cells turn into

Doron Levy Montréal, May 2013

Doron Levy Montréal, May 2013

Doron Levy Montréal, May 2013

Problems with existing therapies

Remission vs. Cure: Can CML be cured?

Doron Levy Montréal, May 2013

The observed immune response

A di↵erent immune response for each patient. However:

Doron Levy Montréal, May 2013

Leukemic ay (a’y ) Progenitor b y (b’y ) Differentiated c y (c’y ) Terminal p0e-cnCkC

d0 + qCp(C,T) d1 + qCp(C,T) d2 + qCp(C,T) d3 + qCp(C,T) dT

Doron Levy Montréal, May 2013

Michor’s model + immune response

Cells without mutations:

ẏ0 = [ry (1 u) d0 ]y0 qc p(C , T )y0 ,

Ṫ = st dt T p(C , T )C + 2n qT p(Cn⌧ , Tn⌧ )Cn⌧ ,

Doron Levy Montréal, May 2013

Accounting for the immune response

Doron Levy Montréal, May 2013

Stopping imatinib (simulation)

Doron Levy Montréal, May 2013

Cancer vaccines: a mathematical design

0.05 -15 -10

Inactivated leukemia cells

Ṫ = st dt T p(C , T )(C + V ) + 2n p(Cn⌧ Tn⌧ )(qT Cn⌧ + Vn⌧ )

Doron Levy Montréal, May 2013

Everything takes time

Doron Levy Montréal, May 2013

Ignore pT2C/C pT2D /TC Survive

Doron Levy Montréal, May 2013

React pT1H /C kCTH kTDTH Perish

Doron Levy Montréal, May 2013

React pT1D/TC kTCTD kTHTD Perish React

Doron Levy Montréal, May 2013

General Donor and Host Blood Cells

Stem Cells Stem Cells

Doron Levy Montréal, May 2013

Doron Levy Montréal, May 2013

Doron Levy Montréal, May 2013

Time for reactive T cell-antigen interaction = 5min

Doron Levy Montréal, May 2013

Cell Concentration in 10 3 cells /µL

Doron Levy Montréal, May 2013

Cell Concentration in 10 3 cells /µL

Doron Levy Montréal, May 2013

A: Stable oscillation B: Unstable Oscillation

Cell Concentration in 10 3 cells /µL

#ROSSINGSET7;SQRT SQRT SQRTSQRT =^;=

REPRESENTSTABLEREGIONS " "#