Epidemiology of
Chronic Myeloid Leukemia
Tom Radivoyevitch, PhD
Assistant Professor
Epidemiology and Biostatistics
Case Western Reserve University
Two CML-ogens: Radiation and Age
3
10
D < 0.02 Sv
0.02 < D < 1 Sv Not exponential => use additive risk model
D > 1 Sv
3
3
3
2
10
2
1
20
30
1
40
Age
Sv = gamma ray dose (Gy)
+ 10 neutron dose (Gy)
50
2
10
Males
k = 0.047
Females k = 0.046
𝑦 = 𝐴𝑒 𝑘∗𝑎𝑔𝑒
10
6
10
10
2
1
2
2
U.S. CML Incidence 1973-2009
Cases per 10 Person-Years
3
6
Cases per 10 Person-Years
Japanese A-Bomb Survivors
60
70
1
0
20
40
Age
60
80
Radiation-induced CML is Multi-scale
For a 500 keV
incoming photon
J = 6.2e18eV
Gy = J/kg
= 6.2e6eV/pL
Figure by
R.K. Sachs.
Stochastic versus Deterministic
Figure by
R.K. Sachs.
Why Study Radiation as the Input?
• Best carcinogen exposure assessment: Abomb survivors remember exactly where
they were, so doses can be reconstructed
• Compared to chemical carcinogen, cannot
simply not use it: background, diagnostic,
and therapeutic exposures are here to stay
• Physics is understood, so results across x& γ-rays, neutrons & protons, and α- and β
particles at different energies can be unified
Other CML-ogen, aging, also cannot be avoided+exposure is known
Why Study CML as the Output?
• CML is homogeneous: all have BCR-ABL
• CML is prevalent: introns large => per-cell
target size for creating bcr-abl is large
• leukemias have rapid onset kinetics: white
blood cells go in and out of tissues naturally
so they don’t need to learn to metastasize
From 1KG browser
Chr 22
49.2 Mb
~5 kb = introns between e12-e15
Chr9 =
136.3 Mb
~140 kb
139.6 Kb DNA Repair 10 (2011) 1131– 1137
PML-RARA intron sizes
40%,55% Mediterr J Hematol Infect Dis. 2011;3(1)
~2kb
~20kb
Seer
APL/CML
1234/10103
= 12%=1/8
40/700=1/18
Dose Response
mi  (e c1  kai  Di ti2 e c2  kt ti ) Pi
mi  [e
c1  kai
 ( Di 


D 
2
i
n

2 c2  kt ti
i
Dni )t e
]Pi e
 ( k Di   k D2i  kn Dni )
NP(ba | T ) w(ti )  ti2 e c2  kt ti
N is the number of CML target
cells in an individual
P(ba|T) is the probability of BCRABL given a translocation
w(t)=probability density that CML
arrives at t given bcr-abl at t=0
kt3t 2 e  kt t
w(t ) 
2

R
2 c k t
 t e 2 t dt 
0
2e c2
kt3
Linear R = 0.0075/Gy. LQE posterior R = 0.0022/Gy
CML Target Cell Numbers
• A comparison of age responses for
CML and total translocations suggests
a CML target cell number of 4x108
• 1012 nucleated marrow cells per adult
and one LTC-IC per 105 marrow cells
suggests 107 CML target cells
• P(ba|T) = 2TablTbcr/2 may not hold
BCR-to-ABL 2D distances
Kozubek et al. (1999) Chromosoma 108: 426-435
Hi-C Data
http://hic.umassmed.edu/heatmap/heatmap.php
chr9
K562 = bcr-abl+ CML cells
133
23
Off by 2 Mb?
chr22
GM06690 = EBV-transformed lymphoblasts
133
23
Lieberman-Aiden, et al. Science
9 October 2009: 289-293.
Theory of Dual Radiation Action

P(ba | D)  2TBCRTABLY D 
2
0
t D (r )
2
S
(
r
)
g
(
r
)
dr


D


D
ba
ba
ba
 4r 2
P(ba|D) = probability of a BCR-ABL translocation per G0/G1 cell given a dose D
tD(r)dr = expected energy at r given an ionization event at the origin
t D (r )  t (r )   4r 2 D =
intra-track component +
inter-track component
Sba(r) = the BCR-to-ABL distance probability density
g(r) = probability that two DSBs misrejoin if they are created r units apart
Y = 0.004 DSBs per Mb per Gy;  = mass density
TBCR = 5.8 kbp; TABL = 140 kbp
Total Translocations → g(r) estimate

2
S 0 (r )  3
2
3
 d
5
r
r
r

(
9
/
4
)

(
3
/
16
)
R3
R4
R6
S0 (r ) ( r / r0 )
1 ( p0G )

t
(
r
)
e
dr

2

4 6.25 0
4r
2

S0 (r ) ( r / r0 )
1 ( p0G )
 dx 
t
(
r
)
e
dr
x
4 6.25 0
4r 2
g (r )  p0 e  ( r / r0 )

1
 d  ( p0G 2 )  S0 (r )e ( r / r0 ) dr
4
0
d in [.01, .025], dx in [.04, .05], d in [.05, .06]
R = 3.7 m  r0 = 0.24 m, p0 = 0.12
G=25 DSB/Gy
6.25 kev/m3 = 1 Gy
Risk and Target Cell Numbers

R   t e
2 c2  k t t
0
2ec2
dt  3  N ba 
kt
N
R
 ba
Dependence of R and N on the choice of fixed LQE parameters ba/ba and ban/ba
BA/BA
.055/.0107
.055/.022
.45/3.64
a
BAn/BA
.8/.0107
.8/.022
.8/.022
R (Gy-1)
.0022 (.0012, .0039)a
.0039 (.0020, .0073)
.0094 (.0051, .0176)
N
6.1x10 (3.3x108, 1.1x109)
5.2x108 (2.7x108, 9.8x108)
7.6x106 (4.1x106, 1.4x107)
8
In parentheses are the 95% CI.
Higher risk estimate is more biologically plausible
Linear-to-quadratic transition dose changed from
[0.011-0.022]/0.055= [0.2-0.4] Gy to
3.64/.45= 8.09 Gy
Linear R = 0.0075/Gy for D < 4Sv is higher
here at 0.0094/Gy due to cell killing term
Bcr-Abl to CML Waiting Times
Eijk  Ai  D j Fk PYijk
Males
Females
kt3t 2 e  kt t
w(t ) 
2
2
10
Males
k = 0.047
Females k = 0.046
10
M/F=1.6
tf-tm=10 yrs
6
4
Cases per 10 Person-Years
6
U.S. CML Incidence 1973-2009
4
Cases per 10 Person-Year-Sv
IR-to-CML Latency
2
M/F=1.42
tf-tm=6.3y
0
1950
1970
Year
1990
1
0
20
40
Age
60
80
mostly radiogenic
Females
mostly radiogenic
5
Cases per 10 Person-Years
0.1
1
10
Males
5
Cases per 10 Person-Years
0.1
1
10
Age at Exposure Dependence
High Dose
Medium Dose
Low Dose
10
20
30
40
Age at exposure
50
High Dose
Medium Dose
Low Dose
10
20
30
40
Age at exposure
50
Nagasaki HSC Reserve Loss?
10
D < 0.02 Sv
0.02 < D < 1 Sv
D > 1 Sv
3
3
3
3
2
10
6
Cases per 10 Person-Years
Japanese A-Bomb Survivors
3
2
1
10
2
1
2
2
10
20
30
6 Nagasaki CML vs 53 in Hiroshima
Hiroshima PY=1558995
Nagasaki PY= 690084 (i.e. 2.26 lower),
53/2.26 = ~23 cases expected in
Nagasaki HSC reserve permanently
depleted to 25%?
1
40
Age
50
60
70
Human T-cell leukemia virus (HTLV): 22
adult T-cell leukemias (ATLs) in Nagasaki
compared to 1 in Hiroshima (2.26 more PY
=> expect ~50)
Dead-Band Control of HSC levels
• Transplant doses of 10, 100, and 1000
CRU => CRU levels 1-20% or 15-60%
normal Blood (1996) 88: 2852-2858
• Broad variation in human HSC levels
Stem Cells (1995) 13: 512-516
• Low levels of HSCs in BMT patients
Blood (1998) 91: 1959-1965
HSC Reserve Loss Trend?
0.80
0.54
6
Males
Cases per 10 PY
2
6
Cases per 10 PY
10
Ave last 7 ratios
0.70
Females
2
0.49 10
10
1
1973-1984 k = 0.058
1985-1996 k = 0.048
1997-2009 k = 0.038
0.1
10
1
1973-1984 k = 0.053
1985-1996 k = 0.049
1997-2009 k = 0.038
0.1
0
20
40
60
Age
80
0
20
40
60
Age
80
1995 data yielded k= 0.041 [Radiat Environ Biophys (1999) 38:201–206].
0.031 in 2006 is consistent with tlcns leading CML by 10 yrs
All Cancer Incidence
Conclusion: Cancer therapy is not the cause of the
HSC reserve depletion
Cumulative Incidence of Cancer
males
females
Other Guesses? Does obesity increase bone
marrow fat and thus squeeze out HSC?
1. Mississippi (34.4%) 51. Colorado (19.8%)
0.1*x+1(1-x)=0.5 => .5=.9x => x=.555
Prevalence of cause must be greater than 55%
0.1
Cancer Epidemiol Biomarkers Prev 2009;18:1501-1506 => obesity
causes CML
2
Easier travel=> greater loads on immune system?
3
Males
10
6
10
2
0
20
40
age
60
80
2
Females
6
10
Cases per 10 PY
10
Cases per 10 PY
probability of cancer
0.5
10
1
1973-1984 k = 0.058
1985-1996 k = 0.048
1997-2009 k = 0.038
0.1
10
1
197319851997-
0.1
0
20
40
60
Age
80
0
20
40
Ag
Or is it CMML Misclassification?
Males
10
CML=ICD9 205.1
includes 20% CMML
10
1
1973-1984 k = 0.058
1985-1996 k = 0.055
1997-2009 k = 0.053
0.1
40
60
Age
10
1
80
1973-1984 k = 0.053
1985-1996 k = 0.054
1997-2009 k = 0.05
0
Males
10
2
20
40
60
Age
80
Females
6
6
2
20
Cases per 10 PY
10
Cases per 10 PY
Females
0.1
0
CML = ICDO-2 9863
does not include CMML.
Maybe all were called
CML <1985, 50% in
1985-1995, and 0 after
2
6
Cases per 10 PY
2
6
Cases per 10 PY
10
10
1
1973-1984 k = 0.058
1985-1996 k = 0.048
1997-2009 k = 0.038
0.1
10
1
1973-1984 k = 0.053
1985-1996 k = 0.049
1997-2009 k = 0.038
0.1
0
20
40
60
Age
80
0
20
40
60
Age
80
CMML rises at older ages
ICDO-2 9945 = CMML
Males
10
2
Females
6
Cases per 10 PY
2
6
Cases per 10 PY
10
10
1985-1996
1997-2009
1
0.1
10
1985-1996
1997-2009
1
0.1
0
20
40
60
Age
80
0
20
40
60
Age
Counts of CMML per year. None before 1985
1984 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
1
40
41
43
50
53
70
61
68
66
65
79
93
65
85
78
82
87
2003 2004 2005 2006 2007 2008 2009
93 127
84
91 104
80 123
80
AML
10
2
AML Females
6
Cases per 10 PY
10
2
6
Cases per 10 PY
AML Males
10
1973-1984
1985-1996
1997-2009
0
20
40
60
Age
10
80
0
20
40
60
Age
6
Cases per 10 PY
APL Females
6
More APL or better
diagnostics?
Cases per 10 PY
APL Males
80
1
1973-1984
1985-1996
1997-2009
0
20
40
60
Age
80
1
0.1
0
20
40
60
Age
80
Retinoic Acid and Imatinib
0.8
0.6
0.0
0.2
0.4
Survival
0.6
0.4
0.2
0.0
40
60
80
100
120
0
50
100
150
200
Months
Months
APL females
CML females
250
300
0.2
0.4
0.6
0.8
1973-1981
1982-1991
1991-1999
2000-2009
0.0
0.0
0.2
0.4
0.6
Survival
0.8
1.0
20
1.0
0
Survival
Cures found for
cancers that are
molecularly
homogeneous:
simpler cancers are
being solved first
Survival
0.8
1.0
CML males
1.0
APL males
0
20
40
60
Months
80
100
120
0
50
100
150
200
Months
250
300
AML and CLL
0.8
0.6
0.0
0.2
0.4
Survival
0.6
0.4
0.2
0.0
100
150
200
250
300
0
50
100
150
200
Months
Months
AML females
CLL females
250
300
0.2
0.4
0.6
0.8
1973-1981
1982-1991
1991-1999
2000-2009
0.0
0.0
0.2
0.4
0.6
Survival
0.8
1.0
50
1.0
0
Survival
More typically
progress is slower
Survival
0.8
1.0
CLL males
1.0
AML males
0
50
100
150
200
Months
250
300
0
50
100
150
200
Months
250
300
Acknowledgements
•
•
•
•
Department of Epidemiology & Biostatistics
Rainer Sachs (UC Berkeley)
Yogen Saunthararajah (Cleveland Clinic)
Thank you for listening!
SEER Underreporting Possibility
Most conservative claims-based algorithm vs. SEER.
B. M. Craig et al. Cancer Epidemiol Biomarkers Prev; 21(3) March 2012
Radiation Doses Rising
AML
10
2
AML Females
6
Cases per 10 PY
10
2
6
Cases per 10 PY
AML Males
10
1973-1984
1985-1996
1997-2009
0
20
40
60
Age
80
10
0
20
40
60
Age
80
Assuming all CML-ogens are also AML-ogens, this implies CML decreases
are NOT due to decreases in exposures to bcr-abl forming agents.
No AML trend is consistent with target cells being lineage committed and
thus more tightly regulated than HSCs.
Others
10
2
6
2
Cases per 10 PY
10
CLL Females
6
Cases per 10 PY
CLL Males
10
1
1973-1984
1985-1996
1997-2008
0.1
0
20
40
60
Age
10
1
80
0
80
Cases per 10 PY
2
10
2
6
10
40
60
Age
MML Females
6
Cases per 10 PY
MML Males
20
10
1
1973-1984
1985-1996
1997-2008
0.1
0
20
40
60
Age
80
10
1
0.1
0
20
40
60
Age
80
All Cancer Incidence
Incidence of All Cancers
10
10
10
Females
2
3
1973-1984
1985-1996
1997-2008
4
0
20
40
Age
60
80
Cases per Person-Year
Cases per Person-Year
Males
10
10
10
2
3
4
0
20
40
Age
60
80
All Cancer Incidence
Cases per Person-Year
Incidence of All Cancers
females
males
10
2
10
3
10
4
2026202 Males
2157740 Females
438616821 MalePY 454528905 FemPY
0
20
40
age
60
80
Cases per 10 Person-Years
0.1
1
10
Nagasaki HSC Reserve Loss?
5
mostly radiogenic
 O log( m )  m
i
i
i
i
Cases per 10 Person-Years
0.1
1
10
53/2.26 = ~23 cases expected in Nagasaki
HSC reserve permanently depleted to 25%?
Human T-cell leukemia virus (HTLV): 22 adult Tcell leukemias (ATLs) in Nagasaki compared to
1 in Hiroshima (2.26 more PY => expect ~40)
10
20
30
40
Age at exposure
50
5
6 Nagasaki CML vs 53 in Hiroshima
Hiroshima PY=1558995
Nagasaki PY= 690084 (i.e. 2.26 lower),
High Dose
Medium Dose
Low Dose
High Dose
Medium Dose
Low Dose
10
20
30
40
Age at exposure
50