Upload
erik-jackson
View
216
Download
0
Tags:
Embed Size (px)
Citation preview
Radiation effects on cancer risks in atomic bomb survivors
Chelyabinsk October 2, 2012
Dale L. PrestonHirosoft International
Eureka, CA
2
OutlineOutline
• Historical background of atomic bomb survivor studies background
• Outline of major survivor cohorts• Dosimetry• Solid cancer risk estimates• Leukemia risk estimates• (Informal) comparison of survivor and Techa River cohort
risks
3
A-bomb Survivor StudiesA-bomb Survivor StudiesAtomicAtomicbombingsbombings
NationalNationalCensusCensus
CohortsCohortsEstablishedEstablished
TumorTumor RegistriesRegistries
1945 1947 1950 1955 1958 1968 1975 1987 1995 20041945 1947 1950 1955 1958 1968 1975 1987 1995 2004
ABCCABCC StudiesStudies
Cancer Incidence Study (LSS)Cancer Incidence Study (LSS)
Clinical Study (AHS)Clinical Study (AHS)
MortalityMortality StudyStudy (LSS)(LSS)
FrancisFrancisComm.Comm.
T65DT65D
RERFRERFcreatedcreated
DS86DS86
BlueBlue RibbonRibbon CommComm..
F1 Mortality StudyF1 Mortality Study
DS02DS02
F1 ClinicalF1 ClinicalStudy (FOCS)Study (FOCS)
JointJointCommComm.
4
Early StudiesEarly Studies
• Genetic effects (malformations, premature birth, sex-ratio)
– 72,000 registered pregnancies 1948 -1953
– No apparent radiation effects
• Leukemia– Initial reports from Japanese physicians in late 1940’s
– First published report in 1952
– Initial crude risk estimates in 1956 (not from ABCC)
• Solid Cancer– First studies in late 1950’s and early 1960’s– Some indication of effects
• Hampered by crude dose estimates and limitations of statistical methods
5
Saving the Survivor StudiesSaving the Survivor Studies
• Calls for end to ABCC studies in early 1950’s– Major genetic studies were completed with
no compelling evidence of hereditary effects– Leukemia excess risk appeared to be declining– Studies being carried out in ad-hoc manner– Costs for program rising– Staff morale low
6
Saving the Survivor StudiesSaving the Survivor Studies
• US National Academies of Science organized committee to assess what should be done about ABCC research
• Recommendations– Reorganized program should continue– Unified study plan
• Focus on fixed cohorts of survivors and their children with internalcomparison groups
• Mortality follow-up
• Pathology (autopsy) program
• Clinical studies
• Highlighted need for dose estimates
7
ABCC/RERF CohortsABCC/RERF CohortsLife Span Study (LSS)Life Span Study (LSS)
A-bomb SurvivorsA-bomb Survivors284,000284,000
Master SampleMaster Sample195195,,000000
Life Span StudyLife Span Study121,320121,320
19501950CensusCensus
1958-1958-
1958-1958-Adult Health StudyAdult Health Study22,00022,000
Original LSS includes groups of non-military Japanese for whom follow-up data could readily be obtained:
1) All survivors' < 2 km with acute effects
2) Matched group of other survivors < 2 km
3) Matched group of people who were 2.5-10km
4) Matched group of unexposed (not-in-city) individuals
8
ABCC/RERF - F1 study cohortsABCC/RERF - F1 study cohorts
F1 MortalityF1 Mortality80,00080,000
Untoward pregnancy Untoward pregnancy outcomesoutcomes
77,00077,000
Biochemical Biochemical Genetic studiesGenetic studies
28,00028,000
FOCSFOCS25,000 selected,25,000 selected, 12,000 examined12,000 examined
Born between Born between May 1946 and May 1946 and December 1984December 1984
BornBorn between between 1947 and 19531947 and 1953
9
ABCC-RERF cohortsABCC-RERF cohortsIn-utero cohortIn-utero cohort
Pooled IU cohort3,638 people
• Pooled cohort combines overlapping clinical (1,606 members) and mortality (2,802 members) cohorts.
• Mortality and cancer incidence data are available for all members of the cohort.
10
ABCC/RERF Follow-up ProgramsABCC/RERF Follow-up Programs
• Mortality– Based on mandatory nation-wide family registration – Updated on a three-year cycle
• Cancer incidence – Hiroshima & Nagasaki tumor registries (1958 – present)– ABCC pathology program 1958 – 1972– Hiroshima & Nagasaki tissue registries 1973 - present
• Leukemia and related disorders– Leukemia registry 1950 – 1987– Hiroshima & Nagasaki Tumor Registries 1958 – present
• Clinical Examinations– Biennial exams– 70-80% participation through 25 AHS exam cycles– Adapted for use in F1 clinical study (FOCS)
• Mail Surveys– 1965 (Ni-hon-san study men), 1968 (women), 1978, 1991, 201?
11
DosimetryDosimetry
• Location– Specified as coordinates on fairly crude US army maps
• Sought corroboration of location
• Recorded to nearest 10m in each coordinate if detailed shielding history obtained and nearest 100m for others
• External Shielding– Crude shielding category information available on virtually all
people of interest– Detailed shielding histories for most survivors within 1.6km in
Hiroshima and 2 km in Nagasaki
• Self shielding (organ dose)– Available for survivors with detailed shielding histories
12
Evolving DosimetryEvolving Dosimetry
• Distance and occurrence of acute effects • “Air dose” curves with crude shielding
adjustments (T57D)• Gamma and neutron air kerma equations with external
shielding models (T65D)– Based on weapons tests with limited validation from
measurements (TLD and Co60 activation)
• Monte-Carlo transport codes for transport and shielding including organ doses (DS86)– More physical measurements
• Updated transport and shielding models (DS02)– Extensive validation efforts especially for neutrons
13
HypocenterDose (mSv)
● < 5 ● 5 – 100 ● 100 – 200 ● 200 - 500
● 500 – 1000 ● 1000 + ▲ unknown * LSS: Life Span Study Cohort
LSS Survivors within 3 KmLSS Survivors within 3 Km
Hiroshima Nagasaki
14
Describing Excess RisksDescribing Excess Risks
Baseline (zero dose) risk function a age at risk; s gender; and b birth cohort
Dose-response shape , e.g. linear, linear-quadratic, threshold, …
Effect modification function e age at exposure
( )d
( , , )a s b
( , , )s e a
Excess relative risk (ERR) model
Excess absolute rate (EAR) model
( , , )[1 ( ) ( , , )]o Ra s b d s e a
( , , ) ( ) ( , , )o Aa s b d s e a
15
LSS Solid Cancer IncidenceLSS Solid Cancer Incidence1958-941958-94
By age at exposure
Age at exposure
People Person years CasesEstimated
Excess AR%*
0-19 21,571 632,341 2,409 150 13%20-39 8,522 229,518 2,569 86 8%40+ 12,809 178,419 2,991 61 5%
Total 42,902 1,040,278 7,969 297 9%
0-19 24,169 755,387 2,186 240 24%20-39 21,561 679,452 4,423 233 11%40+ 16,795 289,614 2,870 83 6%
Total 62,525 1,724,453 9,479 556 13%Total 105,427 2,764,731 17,448 853 11%
By colon doseColon Dose
People Person years CasesEstimated
ExcessAR%
< 0.005 60,792 1,598,944 9,597 3 0% - 0.1 27,789 729,603 4,406 81 2% - 0.2 5,527 145,925 968 75 8% - 0.5 5,935 153,886 1,144 179 16% - 1 3,173 81,251 688 206 30% - 2 1,647 41,412 460 196 43%2+ 564 13,711 185 111 60%
Total 105,427 2,764,732 17,448 853 11%*
Male
Female
• Information on gender and age-time patterns depends (only) on radiation-associated (“excess”) cases
• Excess cases not explicitly identified
• Number of relevant cases is relatively small, especially for specific sites
* Attributable risk % for people with doses > 0.005 Gy
LSS Solid Cancer IncidenceLSS Solid Cancer IncidenceDose ResponseDose Response
• Linear ERR/Gy0 – 2 Gy 0.49
• No evidence of non-linearity (LQ model on 0 – 2 Gy) P > 0.5
• Curvature 0.12LSS-LDEF 1.14
16
Using RERF public dataset lssinc07.csv (www.rerf.or.jp)
Proximal zero dose baseline (adjusted for distal and NIC)
0
.5
1
1.5
exce
ss r
ela
tive
risk
0 1 2 3weighted colon dose (Gy)
Solid Cancer Incidence Dose Response
LSS Solid Cancer IncidenceLSS Solid Cancer IncidenceDose Response 0 – 0.5 GyDose Response 0 – 0.5 Gy
• Linear ERR/Gy0 – 2 Gy 0.490 – 0.1 Gy 0.510 – 0.15 Gy 0.51
• Test for trend0 – 0.1 Gy P = 0.080 – 0.15 Gy P = 0.01
17
Using RERF public dataset lssinc07.csv (www.rerf.or.jp)
Proximal zero dose baseline (adjusted for distal and NIC)
0
.05
.1
.15
.2
.25
exce
ss r
ela
tive
risk
0 .1 .2 .3 .4 .5weighted colon dose (Gy)
Solid Cancer Incidence Dose Response
18
LSS Solid Cancer LSS Solid Cancer Excess Relative Risk Temporal PatternsExcess Relative Risk Temporal Patterns
Age at exposure
-29% per decade(90% CI -39%; -18%)
Attained age
Age-0.9
(90% CI -1.5; -0.2)
Gender *
M: 0.29 (90% CI 0.21; 0.39)
F: 0.58 (90% CI 0.42; 0.68)
F:M: 1.9 (90% CI 1.4; 2.7)
* ERR per Sv at age 70 following exposure at age 30
0 - 9
10 - 19
20 - 39
40+0
1
2
3
4
ER
R p
er S
V
20 40 60 80Age at diagnosis
19
LSS Solid Cancer LSS Solid Cancer Excess Rate Temporal PatternsExcess Rate Temporal Patterns
Age at exposure
-20% per decade(90% CI -30%; -10%)
Attained age
Age3.5
(90% CI 2.9; 4.1)
Gender *
M: 26 (90% CI 18; 34)
F: 28 (90% CI 23; 34)
F:M: 1.1 (90% CI 0.8; 1.6)
* Excess cases per 10000 PY at age 70 following exposure at age 30
0 - 9
10 - 19
20 - 39
40+
0
10
20
30
40
50
exce
ss c
ase
s pe
r 10
,00
0 P
YS
v
20 40 60 80Age at diagnosis
Leukemia Incidence Leukemia Incidence 1950-20011950-2001
• CLL is much less common in Japan than in Russia (or other countries)
• ATL cases mostly in Nagasaki with no indication of a dose response
Leukemia other than CLL or ATL 312
Acute myeloid (AML) ‡ 176
Chronic myeloid (CML) 75
Acute lymphoblastic (ALL) § 43
Other 18
Chronic lymphocytic (CLL) †† 12
Adult T-cell (ATL) 47
Any leukemia 371
Type of Leukemia Eligible
LSS Leukemia IncidenceLSS Leukemia Incidence
21
• Almost half of cases among exposed associated with exposure
Person Observed
years cases Background Excess
< 0.005 2,039,093 0.0006 120 116.9 0.1
-0.1 957,889 0.03 63 60.7 3.6
-0.2 201,935 0.14 16 13.7 4.1
-0.5 206,749 0.32 25 13.6 11.1
-1 117,855 0.71 24 7.5 18.2
-2 64,122 1.37 35 4 28.4
2+ 25,761 2.68 29 1.5 28.6
Total 3,613,404 0.1 312 217.9 94.1
Dose (Gy)
Mean dose (Gy)
Fitted cases
Leukemia IncidenceLeukemia Incidence
• Significant non-linearity in dose response• Low dose risks about half of linear model estimates
22
0
2
4
6
8
10
exce
ss r
elat
ive
ris
k
0 1 2 3weighted dose (Gy)
Leukemia IncidenceLeukemia IncidenceTemporal PatternsTemporal Patterns
23
10
50
30
0
2
4
6
8
exce
ss r
elat
ive
risk
20 40 60 80attained age
ERR
any age at exposure0-19 20 -39
40+
0
1
2
3
4
exce
ss c
ases
per
10,
000P
Y
20 40 60 80attained age
EAR
• ERR model with age and age at exposure dependence describes data as well as EAR model
• EAR model with only attained age effect fits better than age at exposure and time since exposure model from 1994 (gray curves)
Informal Comparison of LSS and Informal Comparison of LSS and Techa River RisksTecha River Risks
• Solid cancer– ERR risk estimates similar around age 60– Temporal patterns not significant in TRC but estimates exhibit
somewhat different patterns than in LSS (increasing with attained age or age at exposure)
– TRC has very limited power to detect effect modifications
• Leukemia– ERR risk estimates for non-CLL similar with similar temporal
patterns
24
25
Summary and ConclusionsSummary and Conclusions
• Accumulating data and modern analytical methods make it possible to investigate radiation effect modification in some detail
• Data are limited even in the largest cohort– Especially true when examining a site-specific risks and modeling interactions
• Both ERR and EAR descriptions provide equally important and complementary information – Attained age is an important factor in both
– Generalization of age at exposure and gender effects can be difficult
• Pooled analyses may be useful in looking at effect modification– But LSS results may dominate such comparisons
26
AcknowledgmentsAcknowledgments
• We stand on the shoulders of giantsGil Beebe, Seymour Jablon, Jim Neel, Jack Schull
• ABCC/RERF scientists and staff who made the ideas a realityGeorge Darling, Howard Hamilton, Tetsuo Imada, Hiroo Kato, M. Kanemitsu, Bob Miller, Kenji Omae, Itsuzo Shigematsu and hundreds more
• Collaborators Japan
Akio Awa, Harry Cullings, Saeko Fujiwara, Shochiro Fujita, Sachiyo Funamoto, Kyoji Furukawa, Kazunori Kodama, Charles Land, Kiyo Mabuchi, Nori Nakamura, Don Pierce, Elaine Ron, Yukiko Shimizu, Michiko Yamada
URCRM / NCI / JCCRER
Lyudimila Krestinina, Marina Degteva, Alexander Akleyev, Sara Schonfeld, Elaine Ron, Faye Davis